Enterprise Modelling and Information Systems Architectures

Volume 9 No. 1 June 2014

Special Issue on A Roadmap for Business Informatics

Enterprise Modelling and

Information Systems ArchitecturesAn International Electronic Journal

Enterprise Modelling and Information Systems ArchitecturesVol. 9, No. 1, June 2014Table of Contents 1

Table of Contents

Editorial Preface 2

Henderik A. Proper and Marc M.Lankhorst

5 Enterprise Architecture – Towards essentialsensemaking

Ulrich Frank 22 Enterprise Modelling: The Next Steps

José Tribolet and Pedro Sousa and ArturCaetano

38 The Role of Enterprise Governance and Cartography inEnterprise Engineering

Jorge L. Sanz 50 Enabling Front-Office Transformation and CustomerExperience through Business Process Engineering

Eng K. Chew 70 Service Innovation for the Digital World

Stéphane Marchand-Maillet and BirgitHofreiter

90 Big Data Management and Analysis for BusinessInformatics – A Survey

Thomas Setzer 106 Data-Driven Decisions in Service Engineering andManagement

Imprint 118

Editorial Board 119

Guidelines for Authors 120

Enterprise Modelling and Information Systems ArchitecturesVol. 9, No. 1, June 2014

2 Editorial Preface

Editorial Preface

In 2011 the IEEE Technical Committee on Elec-tronic Commerce decided to broaden its scopeand, accordingly, rename itself to the IEEE Tech-nical Committee on Business Informatics andSystems. In line with this change in name andscope it decided to rename its flag ship confer-ence to IEEE Conference on Business Informat-ics (CBI). Following these changes, it has beena first priority of the technical committee to ex-actly define the meaning of the term "BusinessInformatics" in an IEEE context and to underpinthe need for a Business Informatics Conferenceunder the umbrella of the IEEE.

Evidently, the IEEE as the Institute of Electricaland Electronics Engineers, the world’s largestprofessional association for the advancement oftechnology, takes a mainly engineering sciencesdirection when approaching Business Informat-ics. In order to find its own scope for the IEEEConference on Business Informatics, we havebeen inspired by Nygaard who defined inform-atics as the science that has as its domain in-formation processes and related phenomena inartefacts, society and nature. In the spirit ofthis definition, we consider Business Informat-ics as a scientific discipline targeting informa-tion processes and related phenomena in a socio-economical context, including companies, organ-izations, administrations and society in general.

A key characteristic of Business Informatics re-search is that it considers a real-world businesscontext in developing new theories and conceptsthat enable new practical applications. Thereby,Business Informatics research does not only ex-tend the body of knowledge of the informationsociety, but at the same time provides a tangibleimpact to industry. Consequently, Business In-formatics is a fertile ground for research with thepotential for immense and tangible impact. Orput it in other words - Business Informatics isresearch that matters!

There is no doubt that Business Informatics isan inter-disciplinary field of study. It endeav-ours taking a systematic and analytic approachin aligning core concepts from management sci-ence, organisational science, economics informa-tion science, and informatics into an integratedengineering science. Consequently, the field ofBusiness Informatics involves a broad spectrumof more specific research domains that focus onimportant aspects of Business Informatics in theabove mentioned context. For the first edition un-der the new title and scope, it has been importantto sharpen the future research directions in thedomain of Business Informatics. Thus, we hadcarefully selected appropriate research domainsthat represent the IEEE understanding of Busi-ness Informatics. In order to reach a common un-derstanding of these domains in our community,we invited distinguished experts to introduce aresearch domain by defining its scope, its exist-ing body of knowledge, and most importantlyits future research challenges. These keynoteshave been a means to guide the community in itsway forward and provide directions for BusinessInformatics in the IEEE CBI context.

In this special issue of the EMISA journal we in-clude seven papers, each based on a IEEE CBI2013 keynote introducing a research domain inBusiness Informatics. Evidently, these papers areneither classical research papers nor pure sur-veys, since they focus to a large extent on the "fu-ture", i.e. the open research challenges (withoutproviding a solution). In the following, we definethe scope of the seven research domains and inparentheses we name the author(s) who intro-duce(s) the domain by a paper presented in thisspecial issue.

1. Enterprise Architecture (Henderik A. Properand Marc M. Lankhorst)Scope: In contrast to partial architectures such asIT architecture or software architecture, enter-prise architecture focuses on the overall enter-prise. Enterprise architecture explicitly incorpor-

Enterprise Modelling and Information Systems ArchitecturesVol. 9, No. 1, June 2014Editorial Preface 3

ates business-related concepts and artefacts inaddition to traditional IS/IT artefacts. By embra-cing an enterprise-wide perspective enterprisearchitecture provides a means for organizationsto coordinate their adaptations to increasinglyfast changing market conditions which impactthe entire enterprise, from business processes toIT support.

2. Enterprise Modelling (Ulrich Frank)Scope: Enterprise modelling is concerned withthe modelling of different aspects of an enter-prise (goals, capabilities, organizational struc-tures, business processes, resources, information,people, constraints, etc.) and their interrelation-ships. Accordingly, enterprise modelling offersdifferent perspectives of an enterprise suitablefor strategic planning, organizational design andsoftware engineering. It covers the notation andsemantics of enterprise modelling languages, theprocesses involved in creating and managingmodels, tool support, as well as quality of model-ling.

3. Enterprise Engineering (Jose Tribolet, PedroSousa, and Artur Caetano)Scope: The enterprise engineering domain aimsto apply an engineering based approach to thedesign of enterprises and their transformation.As such, this domain is concerned with the de-velopment of new, appropriate theories, mod-els, methods and other artefacts for the analysis,design, implementation, and governance of en-terprises by combining (relevant parts of) man-agement and organization science, informationsystems science, and computer science.

4. Business Process Engineering (Jorge Sanz)Scope: Business Informatics deals with informa-tion processes in organizations, industries andsociety at large. This concept of "informationin motion" links to business processes deeply.Processes are the expression of the behaviourof organizations and this behaviour leaves foot-prints in the form of artefacts of all sorts, in-cluding information. Thus, Business Informaticsprofoundly intersects with the social enterprise

from a unique perspective, namely, the integra-tion of information and people’s behaviour.

5. Business (Model) & Service Innovation (EngChew)Scope: Being successful in business no longerdepends on having the "best" product, but in-creasingly depends on delivering high quality ser-vices, through attractive customer-centric busi-ness models, at affordable costs. This forces en-terprises to continuously develop/ innovate theirservices and renew/innovate their business mod-els. The world’s evolution toward services-basedclusters also brings new trends that blur the tradi-tional boundaries across conventional industries,thus generating new opportunities for economiesof scale and scope. This has led to increasing in-terests by disparate industries around the globein the "art and science" of the practices of ser-vice innovation. A new concept, called service-dominant logic, has recently been introduced inthe business discipline to study service phenom-ena - one that has significant cross-disciplinaryimplications for the research and design of IT-enabled service innovations and the attendantservice systems.

6. Empowering & Enabling Technologies (Ste-phane Marchand-Maillet and Birgit Hofreiter)Scope: Enabling technologies in Business Inform-atics integrate management practices with In-formatics and Information Technologies.Business Informatics tasks may be performed,supported or monitored by automated or semi-automated technologies. Running environmentsrange from thin mobile clients to large-scale dis-tributed platforms, and newer areas such as ana-lytics services, big data. Accordingly, we seekpapers for original and innovative empoweringand enabling technologies in domains related toBusiness Informatics.

7. Data-Driven Service and Market Engineering(Thomas Setzer)Scope: Economic problems faced by today’s or-ganizations as well as society as a whole de-mand interdisciplinary knowledge from econom-


4 Editorial Preface

ics, management and informatics. Thus, eco-nomic modelling of IT-based solutions for ana-lytically and statistically formulated economicproblems is subject to this track. In particular,we are interested in the intelligent reduction ofproblem-relevant features from vast datasets In-cluding customer dynamics, market behaviour,resource usage, etc.

It should be noted that these research domainsrepresent cornerstones of the CBI conferenceseries. However, it is our vision to complementthe CBI picture on business informatics by otherappropriate research domains. We plan to intro-duce these domains both at future CBI Keynotesand subsequent special journal issues.

All articles in this EMISA special issue werehanded in by domain experts that have givena keynote presentation at the IEEE Int’l Confer-ence on Business Informatics (CBI 2013), Vienna,15th - 18th July 2013. These invited papers havethen undergone a blind review for EMISA journalpublication. Each paper had been assigned totwo international reviewers. The reviewers foreach papers have been chosen on the followingcriteria: The first reviewer has been a memberof the IEEE CBI steering committee in order toensure compliance of the paper with the scopeof business informatics in an IEEE context. Thesecond reviewer has been an accredited expertin the respective research domain not being in-volved in the IEEE CBI organization in order toensure an open and unbiased representation ofthe domain. In the case where one guest editor isa co-author of the paper, the review process wasmanaged by the other guest editor.

Birgit HofreiterChristian Huemer

Guest Editors’ contact information:

Dr. Birgit HofreiterElectronic Commerce GroupVienna University of TechnologyFavoritenstrasse 9-111040 Vienna, Austria

Prof. Dr. Christian HuemerBusiness Informatics GroupVienna University of TechnologyFavoritenstrasse 9-111040 Vienna, Austria

Enterprise Modelling and Information Systems ArchitecturesVol. 9, No. 1, June 2014Enterprise Architecture 5

Henderik A. Proper and Marc M. Lankhorst

Enterprise ArchitectureTowards essential sensemaking

In this position paper, we discuss our view on the past and future of the domain of enterprise architecture. Wewill do so, by characterising the past, and anticipated future, in terms of a number of trends. Based on thesetrends, we then discuss our current understanding of the future concept and role of enterprise architecture.We conclude by suggesting vantage points for future research in the field of enterprise architecture.

1 Introduction

Increasingly, organisations recognise enterprisearchitecture as an important instrument to steer(or influence) the direction of transformations(Buckl et al. 2011; Greefhorst and Proper 2011;Lahrmann et al. 2010; Lankhorst 2012b; Op ’t Landet al. 2008; The Open Group 2009). Over the pastdecades, the domain of enterprise architecturehas seen a tremendous growth, both in termsof its use and development in practice and as asubject of scientific research. The roots of thedomain can actually be traced back as far as themid 1980s.

In this position paper, which builds on Proper(2012), we will review the evolution of the fieldof enterprise architecture. We do so by charac-terising both its history (Sect. 2), as well as itsanticipated future (Sect. 3), in terms of a numberof trends. Based on these trends, we also dis-cuss our current understanding of the conceptand role of enterprise architecture (Sect. 4). Weconclude with a brief discussion of our view onresearch in the field of enterprise architecture interms of key vantage points for further research.

2 A History of Enterprise Architecture

In this section we discuss the history of the fieldof enterprise architecture in terms of a numberof trends as observed by us.

2.1 From Computer Architecture to ISArchitecture

The origins of enterprise architecture can be tracedback to the concept of information systems ar-chitecture (IS Architecture), which in turn hasits roots in the concept of computer architecture.One of the first references to the term architec-ture, in the context of IT, can be found in a paperfrom 1964 on the architecture of the IBM Sys-tem/360 Amdahl et al. 1964. There it was used tointroduce the notion of computer architecture.

Later, in the 1980s, the term architecture star-ted to become used in the domain of informa-tion systems development as well. This occurredboth in Europe and North America. The NorthAmerican use of the concept of architecture inan information systems context can (at least) betraced back to a report on a large multi clientstudy, the PRISM1 project Hammer & Company(1986) conducted, as well as the later paper byZachman (1987). The European origins can betraced back to the early work of August-WilhelmScheer on the ARIS framework, also dating backto 1986 (Scheer 1986, 1988, 2000).

In Europe, the ARIS framework as developed byAugust-Wilhelm Scheer eventually formed thebase of the well known IDS-Scheer toolset. In

1Not to be confused with the present day concept ofPRISM http://en.wikipedia.org/wiki/PRISM_(surveillance_program)

http://en.wikipedia.org/wiki/PRISM_(surveillance_program)

http://en.wikipedia.org/wiki/PRISM_(surveillance_program)


6 Henderik A. Proper and Marc M. Lankhorst

North America, the PRISM project was a multi-year research project, led by Michael Hammer,Thomas Davenport, and James Champy. PRISM,short for Partnership for Research in Informa-tion Systems Management, was sponsored byapproximately sixty of the largest global compan-ies (DEC, IBM, Xerox, Texaco, Swissair, Johnsonand Johnson, Pacific Bell, AT&T, etc.). This re-search effort produced an architecture frameworkknown as the PRISM Architecture Model, whichwas published in 1986. The PRISM framework hasstrongly influenced other enterprise architecturestandards, methods and frameworks (Beijer andDe Klerk 2010; Davenport et al. 1989; Richardsonet al. 1990; Rivera 2007).

Many years later, the PRISM report also influ-enced the IEEE definition of architecture, as manyof the IEEE 1471 committee members were em-ployed by the original sponsors of their earlierwork on PRISM. Key people involved in PRISMlater also spearheaded the wave on Business Pro-cess Reengineering (Davenport et al. 1989; Ham-mer 1990), which is essentially an early businessarchitecting effort.

The Zachman (1987) paper is often referred to asone of the founding papers of the field of enter-prise architecture. It should be noted, however,that both the PRISM and ARIS frameworks pre-date the Zachman framework, although theseframeworks have indeed been published in lessaccessible sources.

The important message of the ARIS, PRISM andZachman frameworks is the need to considerinformation systems from multiple perspectivesbased on stakes, concerns, as well as different as-pects of the information systems and its businessor technology context, while at the same timefocusing on the key properties of the informa-tion system. The latter focus is also capturedby the phrase fundamental organization in theIEEE 1471 IEEE 2000 architecture definition: “thefundamental organization of a system embodiedin its components, their relationships to each otherand to the environment, and the principles guid-ing its design and evolution.”, where fundamental

is dependent on the key concerns/stakes of thestakeholders involved in an architecting effort.

The basic idea to consider information systems ina holistic way, i.e., from multiple related perspect-ives, was actually already identified before beinglinked to the term information systems architec-ture. For example, Multiview Wood–Harper et al.(1985) already identified five essential viewpointsfor the development of information systems: Hu-man Activity System, Information Modelling,Socio-Technical System, Human-Computer In-terface and the Technical System. Even thoughthe authors of Multiview did not use the termarchitecture, one can argue that Multiview is ef-fectively one of the earliest explicit informationsystems architecture frameworks. During thesame period in which Multiview was developed,the so-called CRIS Task Group of the IFIP work-ing group 8.1 developed similar notions (Olleet al. 1982, 1983), where stakeholder views werecaptured from different perspectives. Special at-tention was paid to disagreement about whichaspect (or perspective) was to dominate the sys-tem design (viz. “process”, “data” or “behaviour”).In the early 1980s, the CRIS Task Group alreadyidentified several human roles (stakeholders!) in-volved in information system development, suchas responsible executive, development coordinator,business analyst, business designer, quite similarto the stakeholder dimension of, e.g., the Zach-man framework.

In the 1990s, challenges such as interoperabilityand distributed computing resulted in the cre-ation of reference architectures, including theCIMOSA (Open System Architecture for CIM)framework for computer integrated manufactur-ing systems ESPRIT Consortium AMICE 1993and the RM-ODP (Reference Model for Open Dis-tributed Processing) framework for informationsystems (ISO 1996a,b, 1998a,b)

2.2 From IS Architecture to EnterpriseArchitecture

The awareness that the design of informationsystems needed to be seen in a broader business


and enterprise context, triggered several authorsto shift towards the use of the term enterprisearchitecture rather than information systems ar-chitecture. One of the first authors to use theterm enterprise architecture was Spewak (1993).

The initial architecture approaches focused onthe development of information systems, whiletaking the models/architectures of other relevantaspects of the enterprise as a given. However,due to the strong connection between businessprocesses and the underlying information sys-tems, it was only natural to not just treat suchperspectives as a given, but rather to co-designthese in tandem with the information systemsand their underlying IT support.

Earlier versions of TOGAF (The Open Group2005), including TAFIM (1996), treated businessarchitecture as a given thing. By defining Enter-prise Architecture Planning (EAP) as “the processof defining architectures for the use of informationin support of the business and the plan for im-plementing those architectures”, Spewak Spewak1993 also seems to suggest to take business archi-tecture as a given. Boar (1999) in “ConstructingBlueprints for Enterprise IT architectures” does thesame.

The shift from taking a business architecture as agiven input, to the realisation that business andIT should be co-designed as a whole, could beseen as the birth of modern day enterprise archi-tecture. The strategic alignment model by Hende-rson and Venkatraman (1993) has played an im-portant role in taking this step to the co-designof business architecture and information systemsarchitecture. Henderson and Venkatraman (1993)indeed suggests that aligning business and ITshould not necessarily require that the businessstrategy should be treated as a given. There areseveral ways to align business and IT. Also thework by, e.g., Tapscott and Caston (1993) contrib-uted to this realisation, as well as the work byRoss et al. (2006). The earlier mentioned work onBusiness Process Reengineering (Davenport et al.1989; Hammer 1990), essentially an early businessarchitecting effort, also contributed to this shift.

Without an attempt to be complete, some enter-prise architecture approaches that indeed take amore co-design oriented perspective include: theIntegrated Architecture Framework (IAF) (Goed-volk et al. 1999; Wout et al. 2010), the ArchiMate(Jonkers et al. 2003; Lankhorst 2012b) language,as well as the DYA (Wagter et al. 2001, 2005) andDEMO (Dietz 2006) methods. Also the most re-cent version of TOGAF (The Open Group 2009)does indeed suggest to co-design the businessarchitecture and the information systems archi-tecture.

2.3 From Business-to-IT-stack toEnterprise Coherence

The realisation that information systems archi-tecture and business architecture need to be co-designed in tandem, led most enterprise archi-tecture approaches to capture a business archi-tecture in terms of building blocks such as busi-ness services, business processes, business actors,etc. These business building blocks were thenlinked to information systems, and ultimatelyIT infrastructures, resulting in a ‘Business-to-IT-stack’. Among an increasing group of researchersand practitioners, the ‘reduction’ of ‘the architec-ture of the enterprise’ to a ‘Business-to-IT-stack’caused unease. In particular Graves (2008), Feh-skens (2008) as well as Wagter (2009) have ar-gued that such a Business-to-IT-stack centricityis a major weakness of contemporary enterprisearchitecture approaches, and that enterprise ar-chitecture should involve many more aspects ofan organisation, such as a clear connection to itsstrategy, its financial structures, the abilities of itswork force, etc. More specifically, Wagter (2009)argue that enterprise architecture should not justbe concerned with Business-IT alignment, butrather with the alignment of all relevant aspectsof an enterprise. Therefore, rather than usingthe term alignment, Wagter (2009) suggest to usethe term enterprise coherence to stress the multi-faceted nature.

A first enterprise architecture method to indeedexplicitly move beyond a Business-to-IT-stack



centricity is the GEA method (Wagter 2009). GEAargues that the coherence between several as-pects of an enterprise needs to be governed ex-plicitly by means of an enterprise architecture.To indeed co-design the different aspects of anenterprise architecture, and to use it (both theco-design process, and the resulting architecture)in governing enterprise coherence, it is necessaryto take the concerns and associated strategic dia-logues of senior management as a starting point.In other words, the way in which architecture isintegrated into the strategic dialogue should takethe concerns, language, and style of communic-ation of senior management as a starting point,and not the typical domains, layers, or columns,as identified in the traditional architecture frame-works.

The shift from Business-to-IT-stack centricity tothe broader notion of enterprise coherence alsorequired a change in perspective on change pro-cesses in organisations (Wagter et al. 2011). DeCaluwé and Vermaak (2003) have identified anumber of core perspectives on change processesin organisations:1: Yellow-print thinking: Bring the interests ofthe most important players together by means ofa process of negotiation enabling consensus or awin-win solution.2: Blue-print thinking: Formulate clear goals andresults, then design rationally a systematic ap-proach and then implement the approach accord-ing to plan.3: Red-print thinking: Motivate and stimulatepeople to perform best they can, contracting andrewarding desired behaviour with the help ofHRM-systems.4: Green-print thinking: Create settings for learn-ing by using interventions, allowing people tobecome more aware and more competent on theirjob.5: White-print thinking: Understand what under-lying patterns drive and block an organisation’sevolution, focusing interventions to create spacefor people’s energy.

As argued in Wagter et al. (2011), most tradi-tional approaches and frameworks, including theSowa and Zachman (1992) and IAF (Wout et al.2010) frameworks, the ArchiMate (Iacob et al.2012; Lankhorst 2012b) language, as well as theDYA (Wagter et al. 2005) and TOGAF (The OpenGroup 2009) architecture methods, essentiallytake a Blue-print perspective on change. Theneed to really involve senior management, how-ever, suggests the use of another style of think-ing, involving internal or external stakeholderinterests, strategy formulation processes, formaland informal power structures, and the associ-ated processes of creating win-win situations andforming coalitions. In terms of De Caluwé andVermaak (2003) this would suggest to comple-ment the Blue-print perspective with the Yellow-print perspective, and arguably also a mix of theother perspectives.

In the development of the GEA method (Wagter2009), this line of thinking was taken as a startingpoint. As a result, the actual political power struc-tures, and associated strategic dialogues, within aspecific enterprise were taken as a starting point,rather than the frameworks suggested by exist-ing architecture approaches. This leads to en-terprise specific frameworks of coherence gov-ernance perspectives, to manage enterprise coher-ence. For example, in terms of ‘mergers & acquis-itions’, ‘human resourcing’, ‘clients’, ‘regulators’,‘culture’, ‘intellectual property’, ‘suppliers’, etc.The existing Blue-print oriented frameworks canstill be used to further structure the dialoguebetween the coherence governance perspectives,especially where it concerns issues pertaining tothe Business-to-IT-stack.

It is to be expected that organisations aiming touse enterprise architecture to steer major trans-formations, will increasingly move from a Busi-ness-to-IT-stack centricity perspective to an en-terprise coherence perspective on their enter-prise architectures.


2.4 From Big-Design-Up-Front toFit-for-Purpose

Early frameworks and languages for enterprisearchitecture (Lankhorst 2012b; The Open Group2005; Wout et al. 2010; Zachman 1987) were primar-ily concerned with the identification of the as-pects, concepts and domains that should be in-cluded in an architecture; hence the resulting con-tent frameworks. This orientation brings alongthe risk that architects focus more on complete-ness of architecture descriptions, rather than onensuring that the descriptions meet the purposesfor which they are actually needed. Acceptedstandards for defining architecture, such as theearlier quoted IEEE 1471 IEEE 2000: “the funda-mental organization of a system embodied in itscomponents, their relationships to each other andto the environment, and the principles guiding itsdesign and evolution.” do not provide a clear‘stop criterion’ for architects that allows them toprovide just enough architecture. This definitionpoints primarily at what the things are that anarchitecture is concerned about: “its components,their relationships to each other and to the envir-onment, and the principles guiding its design andevolution”. The risk is that inexperienced (andmethod obeying) architects loose themselves inmeticulous designs of the future enterprise. Thereference to “the fundamental organization” onlyimplicitly refers to the purpose for having an ar-chitecture, i.e., understanding or expressing thefundamental organisation of a system. But why?And what part of organisation is to be regarded asfundamental? This is of course dependent on thepurpose for which the architecture (description)is created. The more recent ISO (2011) version ofthis definition: “fundamental concepts or proper-ties of a system in its environment embodied in itselements, relationships, and in the principles of itsdesign and evolution”, does not remedy this.

In our observation, the focus on completenessindeed quite often results in overly-detailed ar-chitecture descriptions, involving long lists ofarchitecture principles, meticulously worked outmodels for each of the cells from the architecture

framework used, etc. This situation triggeredthe agile software development community totalk about Ambler and Jeffries (2002); Beck et al.(2001); Cockburn (2002); Lankhorst (2012a) “Big-Design-Up-Front” (BDUF). Of course, experiencedarchitects knew when to stop architecting. How-ever, early architectural approaches did not provideclear guidelines to ensure that architectures stayedFit-for-Purpose, and rather invited architects tobe over-complete.

The need to tune an enterprise architecture tothe purpose at hand and avoid overly detailed ar-chitectures, triggered the authors of Wagter et al.(2001, 2005) to create the DYnamic Enterprise Ar-chitecture approach, which incorporates notionssuch as “just enough architecture”, resembling theideas that were also put forward (in parallel) bythe agile system development community. Themost recent version of TOGAF (The Open Group2009) also provides indications for different (pur-pose specific) ways to use its ADM to ensurethe resulting architecture descriptions are indeedfit-for-purpose.

Greefhorst and Proper (2011) suggest to make aclear distinction between:1: The purpose that an enterprise architectureserves. For example, to understand (make senseof) the current/past situation of an enterprise interms of its fundamental properties and concepts,to articulate and motivate (make sense of) a de-sired future situation in terms of fundamentalproperties and concepts.2: The meaning of an enterprise architecture asan artefact. For example, to express (for somepurpose) the fundamental properties and/or con-cepts that underly the present structure of anenterprise, or to express the fundamental prop-erties and/or concepts that should inspire, guide,or steer, the evolution towards the future.3: The elements of an enterprise architecture interms of the typical concepts used to capture thismeaning, such as its elements, relationships, andthe principles of its design and evolution as men-tioned by the IEEE and ISO definitions, whichmay be captured by means of models and views.



This distinction enables a clear top-down reas-oning on the level of detail and completenessneeded from an architecture description. Giventhe purpose of a specific architecture (descrip-tion), one can identify the desired meaning ofthe architecture, and following that, the kinds ofelements needed to capture/express this mean-ing. For example, Greefhorst and Proper (2011)focus on using enterprise architecture for thepurpose to align the enterprise to its essentialrequirements and ultimately its strategy: “... themain purpose of an enterprise architecture is toalign an enterprise to its essential requirements.As such, it should provide an elaboration of anenterprise’s strategy to those properties that are ne-cessary and sufficient to meet these requirements”.Even though it is only normative in nature, the“necessary and sufficient” and the reference to theenterprise’s strategy provide a (possible) stop-ping criterion to keep an architecture Fit-for-Purpose (i.e., steering transformations that aimto establish an enterprise’s strategy changes).

2.5 From a Constructing to aConstraining Perspective

The shift from computer architecture to inform-ation systems architecture, and then to enter-prise architecture at large, also resulted in anincrease of scope of architecture efforts. Whereat the start the focus was typically on a limitednumber of applications in support of an informa-tion system, the organisational scope graduallybroadened to business-unit wide, then to enter-prise wide, or sometimes even to a sector/branchwide scope. At the same time, the potential time-horizon for architectures increased, from focus-ing on the situation after the next developmentstage, to mid-term and longer-term planningactivities covering several intermediary stages.The shift from Business-to-IT-stack centricity tomore overall enterprise coherence also resultedin a wider range of aspects to be covered in anarchitecture.

The resulting increase in scope and complexity,combined with the Big-Design-Up-Front to Fit-for-Purpose trend as discussed in the previous

section, resulted in the awareness that anothermeans was needed next to the traditional archi-tecture descriptions involving the enterprise’sconstruction in terms of actual building blocks(value exchanges, transactions, business processes,actors, objects, roles, collaborations, etc). Thisresulted in a strengthening of the role of archi-tecture principles as a way to translate an enter-prise’s strategic intentions to more specific dir-ecting/guiding statements, without immediately‘jumping’ to the use of actual building blocksof an actual (high level) design. Several archi-tecture approaches indeed position architectureprinciples as an important element of enterprisearchitecture (Beijer and De Klerk 2010; Daven-port et al. 1989; Op ’t Land et al. 2008; Richardsonet al. 1990; Tapscott and Caston 1993; The OpenGroup 2009; Wagter et al. 2005; Wout et al. 2010),while some authors even go as far as to positionprinciples as being the essence of architecture(Dietz 2008; Fehskens 2010; Hammer & Company1986; Hoogervorst 2009). In our view, the chal-lenges of dealing with increased scope and com-plexity really emancipated the role of principlesas ways to constrain design space.

Fundamentally, we can see a shift from consider-ing an architecture as being primarily concernedwith constructing the (high level) design of anenterprise in terms of building blocks to beingconcerned with constraining the space of allow-able/desirable constructions. A prime exampleof an architecture from a constraining point ofview is the NORA (Nederlandse Overheid Ref-erentie Architectuur (NORA) 2012) reference ar-chitecture for the Dutch government. It focusesprimarily on architecture principles that shouldbe applied in the elaboration of more specificarchitectures and designs.

It is important to note that the distinction betweenconstructing an assembly of building blocks andconstraining the set of possible assemblies to anallowable/desirable subset, is orthogonal to thedeontic modality2 of an architectural description.

2See for example http://en.wikipedia.org/wiki/Deontic_modality

http://en.wikipedia.org/wiki/Deontic_modality

http://en.wikipedia.org/wiki/Deontic_modality


This refers to the question if the architecturaldescription is, for example, intended to be a sug-gestion (could), guidance (advisable), indicative(should), or a pure directive (must).

2.6 From Building to IntegratingAnother trend also resulted in a similar shift to-wards to the constraining of design space. In-stead of developing their own software, most or-ganisations today use packaged solutions, cloudservices and other pre-defined solutions to sup-port large parts of their business activities. Thesesolutions may be configured with the organisa-tion’s business rules, business processes, inform-ation models, etc., but they inherently limit thedesign freedom of the architect. The upside, ofcourse, is in the common gains of re-use: lower-ing costs and risks, and speeding up develop-ment.

This trend, combined with the growing scope andcomplexity outlined in the previous section, alsoleads to a growing emphasis on the integrationof various business processes and IT compon-ents, within and across organisations. Anyonewho has spent some time in a large organisa-tion will recognise that the most common andat the same time most pernicious problems inarchitecture are at these integration points. Theservice-oriented architecture (SOA) paradigm Erl2005 was an important attempt to alleviate thisproblem, but has not been the panacea that itwas once thought to be.

This shift towards integration also influences thedesign and development process. Whereas in thepast, a large system was often designed in one goand as a single, coherent whole, an integrativeapproach will need to be more piecemeal anditerative: adding and integrating various com-ponents one-by-one.

2.7 From One-shot to IterativeApproaches

The agile movement in software development(Ambler and Jeffries 2002; Cockburn 2002) has re-ceived much attention over the last two decades.

Light-weight, iterative methods have graduallytaken over much of the software developmentcommunity. Since the 1990s, evidence has beenmounting that agile ways of working, using shortiterations and close customer contact, have ahigher success rate than traditional, waterfall-like methods for software development, at leastfor many types of software projects. Recent stud-ies provide theoretical and empirical evidencefor the effectiveness of agile methods; see forexample the extensive overview by Lee and Xia(2010).

The Agile Manifesto values “responding to changeover following a plan” (Beck et al. 2001). Manyproponents of agile methods are opposed to theuse of architecture, categorically classifying it asBig-Design-Up-Front. They argue that stakehold-ers cannot know what they really need and theproblem will change anyway before the projectis completed, so one cannot provide any usefuldesigns up-front. Moreover, the changing busi-ness environment makes stable requirements anillusion to begin with. Hence, complex socio-technical systems cannot be designed solely be-hind the drawing board.

On the other hand, many architects and man-agers resist the agile movement, arguing thatone should think before planning actions andbuilding systems. They fear a loss of control andclaim that all these agile projects will build theirown silos, resulting in the same fragmentationof IT landscapes that the architecture disciplinepromised to fix.

Both positions are misguided about the role ofarchitecture. A well-defined architecture helpsin positioning new developments within the con-text of the existing processes, IT systems, andother assets of an organisation, and in identifyingnecessary changes. A good architecture and in-frastructure is an up-front investment that makeslater changes easier, faster and cheaper, and agood architectural practice helps an organisationinnovate and change by providing both stabilityand flexibility (Lankhorst 2012a). But this does



not mean that everything should be architectedup-front. As addressed in Sect. 2.4 and Sect. 2.5,a good enterprise architecture is not overly de-tailed and focuses on the essential inspirationand guidance needed to foster enterprise-widecoherence.

Architecture processes in many organisationsstill give the impression that architects shoulddo all the thinking beforehand and software de-velopers and others can only start their workafter the architects are done. Methods like TO-GAF’s ADM (The Open Group 2009) are alsoeasily interpreted in this way. The measurablesuccess of agile methods and related develop-ments such as continuous delivery (Humble 2010)creates an increasing need for the architecturediscipline to follow suit and embrace a more iter-ative way of working, closely tied to the entiredevelopment process and not merely as a startingphase.

The trend towards less detailed and more normat-ive enterprise architecture, as outlined in Sect. 2.4and Sect. 2.5, matches well with this need foran iterative approach. Agile enterprise archi-tects provide assistance to projects to help themfit within the big picture, while refraining fromtoo much and too detailed guidance. Moreover,as Ciborra (1992) argued, bricolage, emergenceand local improvisation, instead of central con-trol and top-down design, may lead to strategicadvantages: the bottom-up evolution of socio-technical systems will lead to something that isdeeply rooted in an enterprise’s organisationalculture, and hence much more difficult to imitateby others. Agile enterprise architects leave roomfor such local, bottom-up improvements and fitthese within the larger scheme of things.

3 Future trends

In this section we discuss the anticipated futureof enterprise architecture in terms of a numberof anticipated trends.

3.1 From IT to IT

In most enterprises the role of IT started withthe ‘automation of administrative work’. In mod-ern day organisations, there continues to be aclear role for IT to automate administrative in-formation processing. However, the use of IThas moved far beyond this. In some situations,IT has given rise to new social structures, andbusiness models. Consider, for example, the de-velopment of social media, the (acclaimed) role oftwitter in time of social unrest, the emergence ofon-line music stores, app-stores, music streamingservices, etc. The advent of ‘big data’ (Hurwitzet al. 2013) is expected to drive such develop-ments even further by allowing IT based systemsto use statistical data to tune their behaviour toobserved and learned trends.

At the same time, IT is becoming firmly embed-ded in existing technological artefacts. The carsin which we drive now contain more lines ofcode than typical banking applications do. Thenext generation of cars will even be able to (par-tially) do the driving for us. The so-called smart(power) grid, is likely to lead to the ‘smartening’of household appliances. Our houses are alreadybeing vacuumed by dedicated robots, while insome cases robots even play a role in the care ofelderly people (Tamura et al. 2004). The militaryuse of all sorts of drones also spearheads morepeaceful applications of such self-reliant devicesthat can, e.g., perform tasks on behalf of us inhostile or unpleasant environments.

In sum, we argue that we are moving towardssmart and more ‘sociable’ technology that is en-abled by computer technology. One might indeedsay, from information technology to intelligenttechnology, i.e., from IT to IT. When architect-ing modern day enterprises, one should treatthese as (evolving) collectives of human actorsand computerised actors, where the latter mightoperate in a pure software world, or might be em-bedded/embodied in other forms of (connected)technology. Needless to say, however, that hu-man actors will always need to remain (socially


and legally) responsible for the actions of thecomputerised actors that operate on their behalf.

3.2 From Syntax to Semantics

The trend towards an increased scope of integ-ration, described in Sect. 2.6, brings its own setof design issues. Although paradigms such asservice orientation promised to facilitate this in-tegration, they function mainly on a syntacticlevel, providing a stack of interconnection stand-ards for software systems.

When the integration scope grows, the associatedsemantic problems grow as well. The informa-tion shared across organisational borders maybe interpreted in ways that were not intendedand do not match with the context in which thisinformation originated. The same holds for thebehavioural semantics of cross-border businessprocesses. The Semantic Web (W3C SemanticWeb Activity 2013) provides some partial solu-tions, but the premise of its methods is the uni-fication of semantics in a single overarching on-tology, basically trying to standardise the mean-ings of information. It is simply not feasible tobuild such ontologies for the size and variety ofreal-world integration problems. Local varietyin semantics cannot be avoided or ‘standardisedaway’, because of the inevitable loss of meaningthis causes.

This problem is exacerbated by the rapidly grow-ing volume, variety and velocity of ‘big data’(Hurwitz et al. 2013), as already mentioned inSect. 3.1. Applying statistical methods will notsuffice to create meaningful interpretations. Thisimplies that novel methods are needed for archi-tecting the semantics of information and beha-viour, taking into account the variety and contextof meaning and the social processes needed tocreate understanding and agreement at differentscales. It is not feasible to provide complete top-down designs for large-scale socio-technical sys-tems, as we have already argued in Sect. 2.7. Theshift from building towards integration (Sect. 2.6),

also puts more emphasis on the need for se-mantic interoperability. Different semantic back-grounds in a multi-organisational setting makethis even more complicated. We need gradual,iterative approaches for coherent and collaborat-ive design, development and deployment of thesesocio-technical systems.

3.3 From State-thinking toIntervention-thinking

We argue that contemporary approach to archi-tecture ‘think’ in terms of as-is and to-be states ofthe enterprise. Some approaches may indeed goas far as identifying several intermediary stagesbetween as-is and to-be, e.g., leading to the con-cept of transition architecture in TOGAF (TheOpen Group 2009) and plateaus in ArchiMate(Iacob et al. 2012; Lankhorst 2012b). What remainscommon, however, is the focus on several statesof the (construction of the) enterprise. This state-oriented thinking might have worked well in thepast when the focus was on architecting an en-terprise’s IT support. However, as soon as otherother aspects are taken into consideration, thestory becomes more complicated.

As soon as non-technological aspects are takeninto consideration, this brings about a shift offocus from technical systems to socio-technicalsystems involving a mix of human and technolo-gical actors. The enterprise and its environment,being socio-technical systems, will evolve outof themselves. People working in an enterprisewill make changes to the ‘design’ of the enter-prise, if only to make the ‘design’ (continue to)work in day-to-day practise. The people makingup the organisation, collectively ‘author’ theirenterprise (Taylor and Van Every 2010).

Even without the use of architecture as a plan-ning instrument, there are likely to be a plethoraof projects and related efforts that will continu-ously change the enterprise in response to ex-ternal and/or internal stimuli. Some of thesechanges might not even be ‘visible’ as projects,as they are based on local initiatives taken within



the operational processes (i.e., actors switchingbetween a role in the operational capability tothe transformation capability).

We argue that a shift is needed from thinkingof enterprise transformations as being a changeof an enterprise from one state (the as-is) to afuture state (the to-be), but rather as primarilybeing an intervention in the natural evolution ofthe enterprise, resulting in a changed course of itsevolution towards a presumably more desirabledirection. So, from an as-is trajectory to a to-betrajectory.

For the focus of an enterprise architecture thiswould lead to an even stronger emphasis of theconstructing to constraining trend as discussed inSect. 2.5, as constraints are more suitable to artic-ulate desired trajectories than specific buildingblocks. Using, e.g., architecture principles enter-prises can distinguish between desirable and lessdesirable directions of its evolution, and fromthat infer interventions that can be undertakento drive, or lure, the natural evolution of theenterprise in the desired direction. These inter-ventions might indeed involve (re-)constructionsof building blocks of the enterprise.

3.4 From Operational Capability toTransformation Capability

In line with the previous trend, an enterprise islikely to evolve continuously. The capabilitiesneeded to change an enterprise are quite differ-ent from the capability needed to run its day-to-day business. The latter capabilities of an enter-prise can be referred to collectively as its opera-tional capability, while the capabilities needed totransform itself are the transformation capability.Teece et al. (1997) stress the need for modern dayorganisations to have a transformation capabil-ity that meet its rapidly changing environment,leading to a highly dynamic transformation cap-ability: “the firm’s ability to integrate, build, andreconfigure internal and external competences toaddress rapidly changing environments”. Teeceet al. (1997) refer to this dynamic transformationcapability as “dynamic capability”.

It is important to realise that the humans in-volved in an enterprise can play a role towardsboth the operational capability and the trans-formation capability simultaneously. For humanbeings this is actually quite natural. While ex-ecuting our daily activities, we typically alsolearn how to do these activities better and/oradapt them to changing needs/circumstances. Inthese cases, we decide to ‘on the fly’ innovate ouroperational capability. In doing so, we (briefly)use our transformation capability. As a con-sequence, it is advised to regard the operationalcapability and transformation capability of an en-terprise as aspect systems and not as sub systems.

When considering an enterprise from an architec-tural perspective, one can of course opt to focusthe architecture efforts on one of these capabil-ities or both. In most cases that we know of, aswell as the illustrating case studies discussed inthe existing architecture approaches, the focusis on architecting the operational capability only.An exception would be enterprises who have cre-ated a so-called development architecture focusingon the way the enterprise will go about devel-oping new information systems. An example isthe development architecture from the Dutch TaxAdministration (Achterberg et al. 2000).

Whether an enterprise’s architecture effort shouldfocus on the operational capability and/or thetransformation capability depends on the enter-prise’s strategy. For example, in terms of theDiscipline of Market Leaders from Treacy andWiersema (1997), it would be logical for enter-prises focusing on:1: operational excellence, that the operational cap-ability requires architecting priority,2: product leadership, that the parts of the trans-formation capability dealing with product/serviceinnovation require architecting priority,3: customer intimacy, that the parts of the op-erational capability and the transformation cap-ability that deal with client interaction requirearchitecting priority.


When indeed also architecting the transforma-tion capability, it is again recommendable to real-ise that the operational and transformation cap-abilities are aspect systems, and that the differentactors (be they human or be they technology)can play roles towards both capabilities simultan-eously.

In recent work on agile service development(Lankhorst 2012a), it was also argued that an agileservices context requires enterprises to movefrom having only an efficient operational capabil-ity to an effective combination of operational andtransformation capabilities. One should focus ondesigning the operational capability in such away that it lends itself to quick changes withingiven boundaries and ambitions, while the trans-formation capability should be designed in sucha way that it can use this built in agility of the op-erational capability to meet anticipated changesin the environment, as well as the ability to takeappropriate actions to transform the operationalcapability when having to meet unanticipatedchanges (in terms of Teece, it would have to bedynamic).

In Lankhorst (2012a) some guidelines are pro-vided on how to balance an architecting effortbetween the transformation and operational cap-abilities. However, more research is needed. Atthe same time, the need for enterprises to be agile,does stress the need to be able to make explicittradeoffs on how to deal with this agility acrossthe two capabilities.

3.5 From Intuition-based toEvidence-based Management

Modern day enterprises need to change in orderto survive. At the same time they need to do so inthe face of an increasing number of regulationson compliance and transparency. Furthermore, aconsiderable part of an enterprise’s shareholdervalue is ‘tied’ up in the needed transformations.As a consequence, the processes needed to trans-form the enterprise become a core business pro-cess themselves, requiring ample managementattention.

In addition, due to the increasing amount ofshareholder value (and/or taxpayer’s money) thatis tied up in such transformations, one can ex-pect that the requirements on the transparencywith which such decision are made, will increase.Would it not be logical for companies that arelisted on the stock market, to also report annuallyon their ability to transform in an effective way?In other words, not just how well their opera-tional capability is able to earn a revenue for itsshareholders, but also how well their transforma-tion capability is able to ensure the continuationof this revenue in a cost-effective way?

In this sense, one can expect that senior manage-ment will increasingly be held responsible (byshareholders, tax payers, and ultimately auditors)for their ability to steer and control transform-ations. Even more, senior management shouldnot only worry about the cost effectiveness ofchange, but also about governance, risk manage-ment, compliance, etc., associated to these trans-formations. Given the earlier discussion on thepurpose of enterprise architecture, and its rolefor informed governance, it shall not be surpris-ing that we take the point of view that enterprisearchitecture would indeed provide a means tosenior management to take more control overthe transformations and the associated decisionmaking on the future of the enterprises for whichthey are responsible. Using enterprise architec-ture, one can more crisply analyse problems inan existing situation, articulate desired directions(using architecture in a prescriptive way), ana-lyse the costs and benefits of different options(using architecture in a more descriptive way),and guard that transformation projects are in-deed moving in the desired direction.

In parallel to this, one can also observe an in-teresting trend in the field of management. Asargued in (Pfeffer and Sutton 2006, 2011), thereis an increasing call for evidence-based manage-ment instead of (yet not fully replacing) intuition-based management. The authors draw an inter-esting analogy to the trend in medicine towards



evidence-based medicine (Evidence-Based Medi-cine 2012), which is defined in Sacket et al. (1996)as: “the conscientious, explicit and judicious useof current best evidence in making decisions aboutthe care of individual patients.”. If you think thatdoctors would always base their diagnose onsound evidence and reasoning, then Pfeffer andSutton 2011 invites us to rethink this.

When considering the promise of evidence-basedmanagement, there is indeed a strong analogyto the potential contribution of enterprise ar-chitecture. Some early examples of how enter-prise architecture can be used for evidence-basedmanagement of enterprise transformation can befound in, e.g., Op ’t Land (2006, 2007); Op ’t Landand Dietz (2008). We indeed argue that enterprisearchitecture can become a leading mechanism inenabling evidence-based management of trans-formations. Or rather, the field of enterprisearchitecture should take upon it as its missionto enable evidence-based management of trans-formations. We explicitly use the word enableto stress the fact that it is senior managementwho have to take the responsibility for making de-cisions based on evidence. It remains their choicenot to take that responsibility, and explain to theshareholders, tax payers and auditors, why theydid not.

4 Redefining Enterprise Architecture

Based on the future trends as identified in theprevious section, we will now revisit our under-standing of enterprise architecture. In line withthe definition provided in Greefhorst and Proper(2011) we regard architecture as essentially be-ing about: “Those properties of an artefact thatare necessary and sufficient to meet its essentialrequirements”. This view is shared by Fehskens(2008), who defines architecture as “those prop-erties of a thing and its environment that are ne-cessary and sufficient for it to be fit for purposefor its mission”. The focus on the properties thatmatter, is also what distinguishes architecturefrom design. It also resonates well with the refer-ence to fundamental organization in the original

IEEE definition (IEEE 2000) and the reference tofundamental concepts in the ISO definition (ISO2011).

The reference to properties that are necessaryand sufficient to meet its essential requirementsdoes indeed introduce a strong form of relativityto architecture: Who/what determines what theessential requirements are? We argue that theseessential requirements follow from the key stake-holders and their core concerns. What concernsthem most about the artefact? In the case of an en-terprise, the essential requirements can be linkeddirectly to the enterprise’s (past/current) strategy,next to other core concerns of the key stakehold-ers (Greefhorst and Proper 2011). As such, weargue that enterprise architecture should firstand foremost be about essential sensemaking inthat it should primarily:1: make sense of the past and future of the enter-prise with regards to the way it has/will meet itsessential requirements as put forward by its corestakeholders and captured in its strategy,2: provide clear motivations/rationalisation, interms of the above essential requirements, aswell as, e.g., constraints, of the trade-offs that un-derly the presence of the elements (e.g., buildingblocks or architecture principles) included in thearchitecture.

In line with this, we argue that the purpose,mean-ing and elements of an enterprise architectureshould evolve:1: Its purpose is (i) to understand the current evol-ution of the enterprise, including its past and itslikely future evolution and (ii) formulate, as wellas motivation/rationalise, the desired future evol-ution and the interventions needed to achievethis.2: Its meaning is that it expresses, in relationto the (current) essential requirements: (i) theunderstanding how the enterprise has evolvedso-far, (ii) what the expected natural evolutionof the enterprise is and (iii) the desired futureevolution of the enterprise and actions needed tochange/strengthen its current evolution.3: Its elements will focus on the fundamental


properties that have played a role in its past evol-ution, as well as its expected/desired future evol-ution. These properties can be expressed from aconstraining perspective in terms of architectureprinciples and/or from a constructing perspectivein terms of the building blocks of the enterprise.

It is important to note that during the evolutionof an enterprise, it is likely that the understand-ing of what the essential requirements are willchange. This means that the boundary betweenwhat was included in the architecture and whatis considered design may also shift over time.For the modelling languages used (be it from aconstructing or a constraining perspective), thismeans that they should better take a broad per-spective focus on enterprise modelling in gen-eral, where what is considered to be “architectur-ally relevant” may shift over time; modelling ap-proaches with a narrow view of what is “proper”architecture may find themselves obsolete beforethey know it.

5 Conclusion

In this position paper we discussed our view onthe history, and the potential future evolution,of the field of enterprise architecture. It is ourfirm belief that enterprise architecture can, andshould, play a crucial role in enabling senior man-agement of enterprises to take their responsib-ility in steering, controlling and/or guiding en-terprise transformations, based on sensemakingand evidence-based insights. It is certainly oneof the driving hypotheses in our work.

We suggest that future research into the enter-prise architecture domain should do so from atleast three important vantage points, that arealso likely to need different types of researchmethodologies:1: An engineering perspective that focuses onstrategies, methods and techniques to provideevidence-based underpinning of the design de-cisions underlying enterprise architectures (bothin the constructing and the constraining sense).2: A modelling perspective focussing the role of

the different models, frameworks, modelling lan-guages, model transformations, and associatedmodelling processes for enterprise architecture.3: A sociological perspective concerned with therole of culture, skills, attitudes, communication,etc, needed/involved during the formulation ofan enterprise architecture, as well as in the inter-vention needed to establish the changes proposedby a future architectural direction.

6 Acknowledgements

This work has been partially supported by theFonds National de la Recherche Luxembourg, viathe ASINE (Architecture-based Service Innovationin Network Enterprises), ACET (ArchitecturebasedCoordination of Enterprise Transformations) andRationalArchitecture projects.

References

Achterberg R. v., Frankema B., Jong-EllenbroekM. d., Molen P. v. d., Proper H., SchutW. (2000) Handleiding SysteemConcept enApplicatieArchitectuur – Startarchitectuur.Technical Report Version 2.0. Dutch TaxationOffice. Last Access: In Dutch

Ambler S., Jeffries R. (2002) Agile Modeling: Ef-fective Practices for Extreme Programmingand the Unified Process. John Wiley & Sons,New York, New York

Amdahl G., Blaauw G., Brooks F. (1964) Ar-chitecture of the IBM System/360. In: IBMJournal of Research and Development

Beck K., Beedle M., Bennekum A. v., CockburnA., Cunningham W., Fowler M., GrenningJ. Highsmith J., Hunt A., Jeffries R., Kern J.,Marick B., Martin R., Mellor S., Schwaber K.,Sutherland J., Thomas D. (2001) Manifesto forAgile Software Development. http://www.agilemanifesto.org. Last Access: Accessed 14June 2013.

Beijer P., De Klerk T. (2010) IT Architecture:Essential Practice for IT Business Solutions.Lulu

http://www.agilemanifesto.org

http://www.agilemanifesto.org



Boar B. (1999) Constructing Blueprints for En-terprise IT architectures. Wiley, New York,New York

Buckl S., Matthes F., Schweda C. (2011) AMethod Base for Enterprise ArchitectureManagement. In: Ralyté J., Mirbel I., De-neckere R. (eds.) Engineering Methods inthe Service-Oriented Context: 4th IFIP WG8.1 Working Conference on Method Confer-ence on Method Engineering, ME 2011, Paris,France, April 20-22, 2011. Springer, Berlin,Germany, pp. 34–48

Ciborra C. (1992) From Thinking to Tinkering:The Grassroots of Strategic Information Sys-tems. In: The Information Society 8, pp. 297–309

Cockburn A. (2002) Agile Software Develop-ment: The Cooperative Game, 2nd. AddisonWesley

Davenport T., Hammer M., Metsisto T. (1989)How executives can shape their company’sinformation systems. In: Harvard BusinessReview 67(2), pp. 130–134

De Caluwé L., Vermaak H. (2003) Learning toChange: A Guide for Organization ChangeAgents. Sage publications, London, UnitedKingdom

Dietz J. (2006) Enterprise Ontology – Theoryand Methodology. Springer, Berlin, Germany

Dietz J. (2008) Architecture – Building strategyinto design. Netherlands Architecture Forum,Academic Service – SDU, The Hague, TheNetherlands http://www.naf.nl

Erl T. (2005) Service-Oriented Architecture(SOA): Concepts, Technology, and Design..Prentice Hall

ESPRIT Consortium AMICE (1993) CIMOSA:Open System Architecture for CIM, 2nd.Springer, Heidelberg, Germany

Evidence-Based Medicine. http://en.wikipedia.org/wiki/Evidence-based_medicine. Last Ac-cess: Last visited 02-10-2012

Fehskens L. (2008) Re-Thinking architecture. In:20th Enterprise Architecture PractitionersConference. The Open Group

Fehskens L. (2010) What the “Architecture” in

“Enterprise Architecture” Ought to Mean. In:Open Group Conference Boston. The OpenGroup

Goedvolk J., Bruin H. d., Rijsenbrij D. (1999)Integrated Architectural Design of Businessand Information Systems. In: Bosch J. (ed.)Proceedings of the Second Nordic Work-shop on Software Architecture (NOSA’99).Research Report 13 Vol. 1999. University ofKarlskrona/Ronneby, Ronneby, Sweden

Graves T. (2008) Real Enterprise Architecture:beyond IT to the whole enterprise. TetradianBooks, Colchester, England, United Kingdomhttp://tetradianbooks.com

Greefhorst D., Proper H. (2011) ArchitecturePrinciples – The Cornerstones of EnterpriseArchitecture. Enterprise Engineering Series.Springer, Berlin, Germany

Hammer & Company (1986) PRISM: Dispersionand Interconnection: Approaches to Distrib-uted Systems Architecture, Final Report. CSCIndex, Inc.. Cambridge MA

Hammer M. (1990) Re–engineering work: don‘tautomate, obliterate. In: Harvard BusinessReview 68(4), pp. 104–112

Henderson J., Venkatraman N. (1993) Strategicalignment: Leveraging information techno-logy for transforming organizations. In: IBMSystems Journal 32(1), pp. 4–16

Hoogervorst J. (2009) Enterprise Governanceand Enterprise Engineering. Springer, Berlin,Germany, Berlin, Germany

Humble J. (2010) Continuous delivery: reliablesoftware releases through build, test, and de-ployment automation. Addison Wesley

Hurwitz J., Nugent A., Halper F., Kaufman M.(2013) Big Data For Dummies. For Dummies.John Wiley & Sons Inc.

Iacob M.-E., Jonkers H., Lankhorst M., Proper H.,Quartel D. (2012) ArchiMate 2.0 Specification.The Open Group

IEEE (2000) Recommended Practice for Architec-tural Description of Software Intensive Sys-tems. IEEE P1471:2000, ISO/IEC 42010:2007.The Architecture Working Group of the Soft-ware Engineering Committee, Standards De-

http://www.naf.nl

http://en.wikipedia.org/wiki/Evidence-based_medicine

http://en.wikipedia.org/wiki/Evidence-based_medicine

http://tetradianbooks.com


partment, IEEE. Piscataway, New JerseyISO (1996a) Information technology – Open Dis-

tributed Processing – Reference model: Ar-chitecture ISO/IEC 10746–3:1996(E)

ISO (1996b) Information technology – OpenDistributed Processing – Reference model:Foundations ISO/IEC 10746–2:1996(E)

ISO (1998a) Information technology – Open Dis-tributed Processing – Reference model: Archi-tectural semantics ISO/IEC 10746–4:1998(E)

ISO (1998b) Information technology – Open Dis-tributed Processing – Reference model: Over-view ISO/IEC 10746–1:1998(E)

ISO (2011) Systems and software engineering -Architecture description is an internationalstandard for architecture descriptions of sys-tems and software.. ISO/IEC 42010. ISO

Jonkers H., Veldhuijzen van Zanten G., BuurenR. v., Arbab F., De Boer F., BonsangueM., Bosma H., Ter Doest H., GroenewegenL., Guillen Scholten J., Hoppenbrouwers S.,Iacob M.-E., Janssen W., Lankhorst M., VanLeeuwen D., Proper H., Stam A., Torre L.v. d. (2003) Towards a Language for Coher-ent Enterprise Architecture Descriptions. In:Steen M., Bryant B. (eds.) 7th IEEE Interna-tional Enterprise Distributed Object Com-puting Conference (EDOC 2003), Brisbane,Queensland, Australia. IEEE, Los Alamitos,California, pp. 28–39

Lahrmann G., Winter R., Fischer M. (2010)Design and Engineering for Situational Trans-formation. In: Harmsen A., Proper H., Schalk-wijk F., Barjis J., Overbeek S. (eds.) Pro-ceedings of the 2nd Working Conference onPractice-driven Research on Enterprise Trans-formation, PRET 2010, Delft, The Netherlands.Lecture Notes in Business Information Pro-cessing Vol. 69. Springer, Berlin, Germany,pp. 1–16

Lankhorst (ed.) Agile Service Development:Combining Adaptive Methods and Flex-ible Solutions. Enterprise Engineering Series.Springer, Berlin, Germany

Lankhorst (ed.) Enterprise Architecture at Work:Modelling, Communication and Analysis, 3rd.

Enterprise Engineering Series. Springer, Ber-lin, Germany

Lee G., Xia W. (2010) Toward Agile: An integ-rated analysis of quantative and qualitativefield data on software development agility.In: MIS Quarterly 34(1), pp. 87–114

Nederlandse Overheid Referentie Architectuur(NORA). http : / / www . e - overheid . nl /onderwerpen/e-overheid/architectuur/nora-familie/nora. Last Access: Last checked: 02-10-2012

Olle T., Sol H., Verrijn–Stuart A. (eds.) Informa-tion Systems Design Methodologies: A Com-parative Review. North–Holland/IFIP WG8.1,Amsterdam, The Netherlands

Olle T., Sol H., Tully C. (eds.) Information Sys-tems Design Methodologies: A feature ana-lysis. North–Holland/IFIP WG8.1, Amster-dam, The Netherlands

Op ’t Land M. (2006) Applying Architecture andOntology to the Splitting and Allying of En-terprises: Problem Definition and ResearchApproach. In: Meersman R., Tari Z., HerreroP. (eds.) On the Move to Meaningful Inter-net Systems 2006: OTM 2006 Workshops -OTM Confederated International Workshopsand Posters, AWESOMe, CAMS, COMINF,IS, KSinBIT, MIOS-CIAO, MONET, OnToCon-tent, ORM, PerSys, OTM Acadamy DoctoralConsortium, RDDS, SWWS, and SebGIS, Pro-ceedings, Part II, Montpellier, France. Lec-ture Notes in Computer Science Vol. 4278.Springer, Berlin, Germany, pp. 1419–1428

Op ’t Land M. (2007) Towards Evidence BasedSplitting of Organizations. In: Ralyté J.,Brinkkemper S., Henderson-Sellers B. (eds.)Proceedings of the IFIP TC8 / WG8.1 WorkingConference on Situational Method Engineer-ing: Fundamentals and Experiences (ME07),Geneva, Switzerland. IFIP Series Vol. 244.Springer, Berlin, Germany, pp. 328–342

Op ’t Land M., Dietz J. (2008) Enterprise onto-logy based splitting and contracting of or-ganizations. In: Proceedings of the 23rd An-nual ACM Symposium on Applied Comput-ing (SAC’08), Fortaleza, Ceará, Brazil

http://www.e-overheid.nl/onderwerpen/e-overheid/architectuur/nora-familie/nora





Op ’t Land M., Proper H., Waage M., Cloo J.,Steghuis C. (2008) Enterprise Architecture –Creating Value by Informed Governance. En-terprise Engineering Series. Springer, Berlin,Germany

Pfeffer J., Sutton R. (2006) Evidence-Based Man-agement. In: Harvard Business Review Lastvisited: 02-10-2012 http://hbr.org/2006/01/evidence-based-management/ar/1

Pfeffer J., Sutton R. (2011) Trust the Evidence,Not Your Instincts. In: New York Times NewYork edition, September 4th, BU8 http://www.nytimes.com/2011/09/04/jobs/04pre.html?_r=2&ref=business&

Proper H. (2012) Enterprise Architecture –Growing up to evidence-based management?In: Baldinger F., Rijsenbrij D. (eds.) Nether-lands Architecture Forum, The Netherlandschap. 8.1, pp. 317–328

Richardson G., Jackson B., Dickson G. (1990)A Principles-Based Enterprise Architecture:Lessons from Texaco and Star Enterprise. In:MIS Quarterly 14(4), pp. 385–403 http://www.jstor.org/stable/249787

Rivera R. (2007) Am I Doing Architecture orDesign Work? In: It Professional 9(6), pp. 46–48

Ross J., Weill P., Robertson D. (2006) Enterprisearchitecture as strategy: creating a founda-tion for business execution. Harvard BusinessSchool Press, Boston, Massachusetts

Sacket D., Rosenberg W., Gray J., Haynes R.,Richardson W. (1996) Evidence based medi-cine: what it is and what it isn’t. In: Brit-ish Medical Journal 312, pp. 71–72 http :/ /www.ncbi . nlm . nih . gov / pmc / articles /PMC2349778/pdf/bmj00524-0009.pdf

Scheer A.-W. (1986) Neue Architektur fürEDV-Systeme zur Produktionsplanungund -steuerung. In German. Institut fürWirtschaftsinformatik im Institut fürEmpirische Wirtschaftsforschung an derUniversität des Saarlandes, Saarbrücken,Germany

Scheer A.-W. (1988) Computer integrated manu-facturing : CIM. Springer, Berlin, Germany

Scheer A.-W. (2000) ARIS – Business ProcessModeling. Springer, Berlin, Germany

Sowa J., Zachman J. (1992) Extending and form-alizing the framework for information sys-tems architecture. In: IBM Systems Journal31(3), pp. 590–616

Spewak S. (1993) Enterprise Architecture Plan-ning: Developing a Blueprint for Data, Ap-plications, and Technology. Wiley, New York,New York

TAFIM (1996) Department of Defence Tech-nical Architecture Framework for Informa-tion Management – Overview. Defence In-formation Systems Agency Center for Stand-ards, United States of America

Tamura T., Yonemitsu S., Itoh A., Oikawa D.,Kawakami A., Higashi Y., Fujimooto T., Na-kajima K. (2004) Is an entertainment robotuseful in the care of elderly people withsevere dementia? In: J Gerontol A Biol SciMed Sci 59(1), pp. 83–85

Tapscott D., Caston A. (1993) Paradigm Shift –The New Promise of Information Technology.McGraw–Hill, New York, New York

Taylor J., Van Every E. (2010) The Situated Or-ganization: Case studies in the pragmatics ofcommunication research. Routledge

Teece D., Pisano G., Shuen A. (1997) DynamicCapabilities and Strategic Management. In:Strategic Management Journal 18(7), pp. 509–533

The Open Group (2005) TOGAF – The OpenGroup Architectural Framework – Version 8.1Enterprise Edition. The Open Group – VanHaren Publishing, Zaltbommel, The Nether-lands

The Open Group (2009) TOGAF Version 9. VanHaren Publishing, Zaltbommel, The Nether-lands

Treacy M., Wiersema F. (1997) The Disciplineof Market Leaders – Choose your customers,narrow your focus, dominate your market.Addison Wesley, Reading, Massachusetts

W3C Semantic Web Activity (2013). http://www.w3.org/2001/sw/. Last Access: Last visited 14-06-2013

http://hbr.org/2006/01/evidence-based-management/ar/1

http://hbr.org/2006/01/evidence-based-management/ar/1

http://www.nytimes.com/2011/09/04/jobs/04pre.html?_r=2&ref=business&



http://www.jstor.org/stable/249787

http://www.jstor.org/stable/249787

http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2349778/pdf/bmj00524-0009.pdf



http://www.w3.org/2001/sw/

http://www.w3.org/2001/sw/


Wagter R. (2009) Sturen op samenhang op basisvan GEA – Permanent en event driven. InDutch. Van Haren Publishing, Zaltbommel,The Netherlands

Wagter R., Berg M. v. d., Luijpers J., Steenber-gen M. v. (2001) DYA: snelheid en samen-hang in business en ICT architectuur. TuteinNolthenius

Wagter R., Berg M. v. d., Luijpers J., SteenbergenM. v. (2005) Dynamic Enterprise Architecture:How to Make It Work. Wiley, New York, NewYork

Wagter R., Proper H., Witte D. (2011) Enter-prise Coherence Assessment. In: Harmsen A.,Grahlmann K., Proper H. (eds.) Proceedingsof the 2rd Working Conference on Practice-driven Research on Enterprise Transform-ation, PRET 2011, Luxembourg-Kirchberg,Luxembourg. Lecture Notes in Business In-formation Processing Vol. 89. Springer, Berlin,Germany, pp. 28–52

Wood–Harper A., Antill L., Avison D. (1985) In-formation Systems Definition: The MultiviewApproach. Blackwell, Oxford, United King-dom

Wout J. v., Waage M., Hartman H., Stahlecker M,Hofman A. (2010) The Integrated Architec-ture Framework Explained. Springer, Berlin,Germany

Zachman J. (1987) A framework for informa-tion systems architecture. In: IBM SystemsJournal 26(3)

Henderik A. Proper

Public Research Centre Henri Tudor29, avenue J.F. KennedyL-1855 LuxembourgLuxembourgandRadboud University NijmegenFaculty of SciencePostbus 9010, 6500GL NijmegenThe [email protected]

Marc M. Lankhorst

BiZZdesignP.O. Box 3217500 AH EnschedeThe [email protected]


22 Ulrich Frank

Ulrich Frank

Enterprise Modelling: The Next Steps

Enterprise modelling is at the core of Information Systems and has been a subject of intensive research forabout two decades. While the current state of the art shows signs of modest maturity, research is still facingsubstantial challenges. On the one hand, they relate to shortcomings of our current knowledge. On the otherhand, they are related to opportunities of enterprise modelling that have not been sufficiently addressed sofar. This paper presents a personal view of future research on enterprise modelling. It includes requests forsolving underestimated problems and proposes additional topics that promise to promote enterprise models asmore versatile tools for collaborative problem solving. In addition to that, the paper presents requests for (re-)organising research on enterprise modelling in order to increase the impact of the field.

1 Introduction

It has been a wide-spread conviction for longthat the complexity of large information systemsrecommends the use of models. Information sys-tems are aimed at representing domains throughdata that is accessible by prospective users. Rep-resenting a–factual or aspired–domain cannot beaccomplished by modelling it directly. Instead,it comprises a twofold abstraction: We perceivea domain primarily through language, which inturn reflects an abstraction over “objective” fea-tures of a domain. At the same time, using aninformation system requires an interface that cor-responds to the language spoken in the targeteddomain. Therefore, the construction of informa-tion systems recommends the design of concep-tual models. They do not only promise to reducecomplexity by abstracting from ever changing pe-culiarities of technical infrastructures; they alsoallow for getting prospective users involved inthe analysis and design process. Exploiting thepotential of information systems will often re-quire reorganising existing patterns of action—sometimes in a radical way. Therefore, analysisand design of information systems should usuallybe done conjointly with analysing and designingthe organisational action system. The conceptionof an enterprise model was developed to addressthis demand. An enterprise model integrates

at least one conceptual model of an organisa-tional action system with at least one conceptualmodel of a corresponding information system.Usually, but not necessarily, the various mod-els that constitute an enterprise model are cre-ated with domain-specific modelling languages(DSML). To emphasise that enterprise models areintended to provide a medium both for fosteringanalysis and design tasks and for communicationacross traditional professional barriers, the term“multi-perspective enterprise model” has beenintroduced (Frank 1994). A multi-perspective en-terprise model is an enterprise model that em-phasises accounting for multiple perspectives.A perspective represents a specific professionalbackground that corresponds to cognitive dis-positions, technical languages, specific goals andcapabilities of prospective users (Frank 2013b).

In recent years, the term “enterprise architec-ture” has gained remarkable attention (Buckl etal. 2010; Land et al. 2009; Lankhorst 2005). Thedifferences between enterprise model and enter-prise architecture are mainly related to the inten-ded audience. Enterprise modelling is aimed atvarious groups of stakeholders that are involvedin planning, implementing, using and maintain-ing information systems. Therefore, enterprisemodels are supposed to offer a variety of cor-responding abstractions. These include models

Enterprise Modelling and Information Systems ArchitecturesVol. 9, No. 1, June 2014Enterprise Modelling: The Next Steps 23

that serve as a foundation of software develop-ment. Therefore, the development of respectiveDSML is a particular characteristic of enterprisemodelling. Different from that, enterprise archi-tecture mainly targets IT management. There-fore, it puts less emphasis on the specificationof DSML. Nevertheless, there is no fundamentaldifference between both approaches. Instead, theabstractions used in enterprise architectures canbe seen as an integral part of more comprehens-ive enterprise models. In Information Systems,enterprise modelling has been a pivotal field ofresearch that has evolved over a period of morethan 20 years (CIMOSA: Open system architec-ture for CIM 1993; Ferstl and Sinz 1998; Group2009; Scheer 1992; Zachman 1987). It has producedvarious modelling frameworks, DSML as well ascorresponding tools. The field has achieved aremarkable degree of maturity which is indicatedby the fact that enterprise modelling is part ofmany IS curricula—even though to different ex-tent. Nevertheless, there is still need for furtherresearch to exploit the potential of enterprisemodelling. In this paper I will point at relevantshortcomings of the current state of the art inorder identify core elements of a future researchagenda.

2 The Need for More Context

At the beginning, approaches to enterprise mod-elling were mainly focussed on developing high-level frameworks to provide a common structureor architecture of an enterprise and its informa-tion system (CIMOSA: Open system architecturefor CIM 1993; Scheer 1992; Zachman 1987). Apartfrom using general purpose modelling languages(GPML) like the ERM and the UML, the devel-opment of DSML was mainly aimed at businessprocess modelling. Later, DSML were created formodelling strategies, organisational structuresor generic resources. In recent years, some ap-proaches have evolved that are aimed at DSMLfor modelling IT infrastructures and IT architec-tures. This focus is remarkable for two reasons.

At first, it represents a renunciation of the ori-ginal approach to enterprise modelling, whichwas aimed at developing information systemsfrom scratch. With respect to the complexity oftoday’s IT infrastructures and the fact that mostorganisations will not develop substantial partsof their information system on their own any-more, this additional focus is certainly reasonable.Secondly, it is not only aimed at supporting thedesign of IT infrastructures that are in line withthe corporate action system, but also at providingan instrument for IT management. Figure 1 illus-trates the representation of an enterprise modelthrough a set of interrelated diagrams that cor-respond to the current state of the art. To ensureintegration, the partial models that are repres-ented by the diagrams should be created withmodelling languages that were specified with thesame meta modelling language and that sharecommon concepts (Frank 2011).

While the abstractions covered by today’s en-terprise modelling methods arguably representrelevant perspectives on an enterprise, a compre-hensive representation of all aspects that may berelevant for analysing, designing and managinga company together with its information sys-tem requires accounting for more context. Whilesuch a demand may look like an exaggeration tosome, it is actually the consequent continuationof current practice: All professional activities ina company are characterised by the use of con-ceptual abstractions, i.e., by a specific technicalterminology and corresponding language games.Reconstructing these terminologies through ad-ditional DSML would not only enable further usescenarios, it would also enrich existing modelswith additional context. Context does not onlyrefer to the topics that are represented in an en-terprise model. It also refers to the context inwhich the development and use of models occur.On the one hand, this kind of context includesspecific methods for enterprise modelling. Onthe other hand, it refers to organisational andmanagerial arrangements to foster an adequatehandling of enterprise models.


24 Ulrich Frank

Goal System Diagram

Object Model

InboundLogistics Operations

Outbound Logistics

Marketing Sales

Services

Sun SPARC

UNIX HP UX

BL: SAP PLM

SAP CO P10 6.0

SAP FI P10 6.0

SAP R&D P20

SAP

Clu

ster OS

DB: ADDS

Scan Engine 5.1

Document Archive

2.0

Dell Power Edge T110

UNIX HP UX

DMS 1.9

PLM 3.7

Un

ix H

P S

erve

r OS

Value Chain Diagram

Organizational Chart

Business Process Diagram

IT Resource Diagram

Increasesust.

sales in region B

1

Sustainable sales

Increase number of

sales agents

4

Number ofsales agents

Keep labor costs stable

3

Labor costs

Increase sustainabl

e sales

1

Sustainable sales

Increase customer satisfactio

n

2

Customer satisfaction

Increasesust. sales in region

A

1

Sustainable sales

Order Management Product B

<Sales>

Order Management

Product C

<Sales>

Technical Support

<IT Service>

Customer Akquisition

<Sales>

Complaint Handling Product

A

<Customer Service>


B

<Customer Service>


C

<Customer Service>

Order Management Product A

<Sales>

Board of Directors

Marketing Finance Accounting Sales

Market Research

Asset Manageme

ntSystems

Private Customers

Product Planning

Corporate Credits

ControllingCorporate Customers

PromotionInvestor Relations

Decisions Support

Sales Forecast

issues1,1

sum() : Decimaltotal() : Decimal

created: Dateno: String

Order

custID: String

Customer

name: Stringid: StringinStock: IntegersalesPrice: MoneyretailPrice: Money

Product

sum() : Decimal

units: Integer

OrderItem

total() : Decimal

issued: Date

Invoice

balance() : Decimal

id: String

Account

created: Dateamount: Decimal

AccountEntry

name: String

Organisation

street: Stringcity: Stringcountry: String

Address

sum() : Decimal

units: Integer

InvoiceItem

0,*

acts asacts as0,1

0,1

0,10,1

loca

ted

at

loca

ted

at 0,*

1,1

1,1

refe

rs t

o 0,*

1,1

age() : Integer

lastName: StringfirstName: Stringborn: Datesex: Gender

Person 1,*

1,1

incl

ud

es

incl

ud

es

1,*

1,1

0,10,*

resu

lts

in

0,*

0,*

0,*1,1

Business Process Map

ABAP x

MW: SAP Netweaver

MDM

MW: SAP ERC ECCSA

P A

pp

licat

ion

Se

rver

Order recieved

< Agent >

Check Delivery Capacity

amount not available

amount available

< Agent >

Inform Customer

< Fleet Manager >

Check Logistics

Order declined

Delivery possible

Delivery at desired date not possible

< Agent >

Confirm Order

< Agent >

Decline Order

Figure 1: Diagrams Representing Example Enterprise Model


2.1 Further Modelling Topics

The variety of topics that are handled in enter-prises is enormous. Among those that have notbeen sufficiently addressed in enterprise mod-elling are products, production processes, pro-jects, markets and logistic. While product mod-elling is an issue on its own, integrating productmodels with enterprise models makes sense forvarious reasons. Products can be very complexand may demand for quick adaptations. At thesame time, developing, producing and handlingproducts relates to various aspects of an enter-prise that are usually part of an enterprise model:goals, business processes, organisational units orsoftware systems. More and more, products com-prise software or are constituted by software. Inaddition to that products are often bundled—withservices and/or other products. Therefore, integ-rating product models with enterprise modelswould enable additional analysis scenarios suchas checking the effect of changing a product onrequired skills and on business processes. Thereare numerous approaches to model productionprocesses. They aim at developing algorithmsand approximation procedures for productionplanning, process scheduling and process control.Integrating respective models with enterprisemodels will often be not trivial, because theyare based on different modelling paradigms. Atthe same time, including elaborate models of pro-duction processes in enterprise models promisesvarious advantages, such as supporting the con-joint analysis of production processes and relatedbusiness processes or generating software forcontrolling production processes from respectivemodels. In an increasing number of organisa-tions, projects play a key role. Integrating projectmodels with enterprise models would supportproject management by providing meaningfullinks to organisational resources. Also, projectmodelling could benefit from existing approachesto business process modelling and would allowto take advantage of similarities between pro-jects. Markets have not been part of enterprisemodels for an apparent reason: They are outside

of an enterprise and are usually not subject ofdesign processes. Nevertheless, markets are ofcrucial importance for successful action in an en-terprise. Furthermore, markets are getting morecomplex and contingent: Often, they expand onan international scale and may be very dynamicin the sense that products are displaced by in-novations or that customer preferences changequickly. Therefore, integrating models of mar-kets with enterprise models promises to gain amore differentiated understanding of relevantmarket forces and to develop a better founda-tion for decision making. Similar to productionprocesses, logistic networks have been subject ofoptimisation efforts for long. The respective mod-els, often designed to satisfy the requirementsof Operations Research methods, are mainly fo-cussed on optimisation with respect to certaingoals. Integrating the respective modelling con-cepts with languages for enterprise modellingwould enable to enrich both enterprise modelsand logistic models with more relevant context.

In addition to traditional topics, organisations areconfronted with new phenomena that may de-mand for appropriate action. They include socialnetworks, virtual enterprises, nomad employeesand many more. Extending enterprise modelswith models of these phenomena would fosteranalysing and handling them. This would, how-ever, require new modelling concepts. Finally,enterprise models can be supplemented with con-cepts that are related to important further aspectsof managerial decision making. These includeaccounting concepts, e.g., specialised cost andbenefit concepts, concepts to design and analyseindicator systems (Strecker et al. 2012) as wellas concepts for modelling organisational know-ledge. Adding these concepts would make enter-prise models and corresponding tools a versatileinstrument for management—both on the oper-ational and strategic level. At the same time, itcould serve as a valuable extension of enterprisesoftware systems (see Sect. 4.1).

Request: To make enterprise models a versatiletool for supporting professional action in organ-


26 Ulrich Frank

isations, research needs to widen the scope ofmodelling by adding further topics that also com-prise concepts to support managerial decisionmaking.

2.2 Method Construction

Modelling languages are an important founda-tion of enterprise modelling, since they providea purposeful structuring of a domain. However,they are not sufficient for designing and using en-terprise models. In addition to languages, thereis need for processes that guide the purposefuldevelopment, interpretation and use of respect-ive models. In other words: There is need formodelling methods. Due to the diversity of pro-jects that can benefit from conceptual models,it is evident that a given set of modelling meth-ods cannot not fit all demands—except for theprice of oversimplification. This insight shiftedthe focus on approaches that guide the develop-ment of customised methods. The only chanceto provide support for the conceptualisation ofa range of methods is to increase the level ofabstraction by searching for essential character-istics shared by all modelling methods. Againstthis background, the emergence of method en-gineering as a new field of research is a reas-onable consequence in two respects: First, itis aimed at rich abstractions that cover a widerange of modelling projects. Second, it makesuse of the same paradigm that it suggests forthe field of conceptual modelling, too: The con-struction of particular methods should followan engineering approach, which—among otherthings—recommends accounting for linguisticrigour, consistency and coherence as well as forthe development of supportive tools. Duringthe last 15 years, a plethora of method engineer-ing approaches—originating mostly in Require-ments Engineering and Software Engineering—has evolved (Brinkkemper 1996; Ralyté et al. 2005,2007). For an overview see Henderson-Sellers andRalyté (2010). Some emphasise the constructionof methods from reusable elements, others focus

on the instantiation of methods from metamod-els, while further approaches are based on acombination of composition and instantiation.It seems that the field has reached a stage of mod-erate maturity, which is also indicated by the spe-cification of a respective ISO standard (ISO/IEC2007).

Nevertheless, there are some aspects that havebeen widely neglected so far. At first, currentapproaches to method engineering are mostlygeneric in the sense that they are not restrictedto particular domains, nor do they account for thepeculiarities of enterprise models. That leavesprospective developers and users of enterprisemodels with the demanding task of adapting gen-eric concepts to the idiosyncrasies of particularorganisations. Second, current approaches tomethod engineering focus on the design of pro-cess models and take the modelling language asgiven. However, the diversity of topics that canbe reasonable subjects of enterprise models mayalso require to adapt or even create modellinglanguages. While a number of tools support thespecification of DSML and the realisation of cor-responding model editors, prospective users canexpect only little guidance with designing a lan-guage that fits its purpose. At the same timethe design of DSML is especially demanding. Of-ten, prospective users will not have an idea ofwhat they might expect from a DSML. As a con-sequence, requirements analysis is a remarkablechallenge. In addition to that, design conflictsneed to be handled. Also, the creation of the con-crete syntax requires a specific competence thatmany language designers do not have. There-fore, there is need for substantiated guidance toreduce the risk of poorly designed modelling lan-guages. Currently, there are only few approachesthat offer respective support (Frank 2013a; Moody2009). Finally, method engineering is often basedon two assumptions: First, a method is an arte-fact that is created through an engineering act.Second, applying the method appropriately ispivotal for successful action. However, with re-spect to successfully using a method, it is not


sufficient to restrict it to its explicit definition,i.e., to take a mere technical perspective. Thisis for two interrelated reasons. First, a methodwill usually not be based on a pure formal spe-cification. Instead, its conceptual and theoreticalfoundation as well as the process description re-quire interpretations that produce some degreeof shared understanding. Second, for a methodto work, it has to become an accepted orientationfor individual and collaborative action. To sum-marise both aspects: A method needs to makesense. From this point of view, a method canbe regarded as a social construction that reflectsestablished patterns of professional action, ideasof professional values and aesthetics, organisa-tional culture, common beliefs as well as indi-vidual interests. Against this background, wecan distinguish between a method as a linguisticartefact, stressing a technical view, and a methodas an actual practice, stressing a more pragmaticor organisational view. Therefore I intention-ally avoid the term “method engineering” andspeak of “method construction” instead. Thisis to express that a method is also constructedby those who use it, because it is shaped by ac-tual interpretations and actions. A method as anartefact could be regarded as input or stimulusto trigger such a process. While for analyticalreasons it may be useful to focus on methodsmainly as linguistic artefacts, such a restrictedview is certainly not sufficient with respect to apragmatic objective such as improving efficiencyand quality of collaborative problem solving inorganisations. The benefit of methods for en-terprise modelling will not only depend on thequalification of the involved stakeholders, butalso on certain aspects of the respective corpor-ate culture. It makes a clear difference, whetherconceptual models are regarded as corporate as-sets or as cost drivers with dubious outcome.

Request: To promote the beneficial developmentand use of enterprise models it is required tosupport the construction of respective modellingmethods that account for both, the conceptualfoundation of designing/customising methods

as linguistic artefacts and additional organisa-tional/managerial measures that promote the ap-propriate use of methods in practice.

3 The Need for More (Re) Use

The remarkable effort that is required to buildelaborate enterprise models makes reuse of mod-els and modelling concepts a pivotal issue forachieving higher productivity. At the same time,reuse can also contribute to model quality, if re-usable artefacts are designed and evaluated withspecific care. In addition to that reusable con-cepts can serve to foster integration of those com-ponents that share them. Approaches to promotereuse have been on the research agenda for long.The idea of reference enterprise models seemsto be especially attractive. However, so far, re-use of enterprise modelling artefacts remainedon a modest level (Fettke and Loos 2007). Thereare various reasons that contributed to this un-satisfactory situation. Two especially importantreasons are related to modelling languages. Onthe one hand, current languages for enterprisemodelling lack concepts that enable reuse. Onthe other hand, the design of reusable DSML isfacing a substantial challenge.

3.1 The Lack of Abstraction in ProcessModelling

Taken the fact that business process modellinghas been a research subject for long, it seems sur-prising that respective modelling languages arerather primitive in the sense that they do not al-low for powerful abstractions. As a consequence,reuse of business process models remains on adramatically poor level. Since business processmodels play a pivotal role within enterprise mod-els, this is a serious shortcoming. The follow-ing scenario illustrates the problem. A companycomprises a few tens of business process typesincluding a core order management process type.A process type includes activity types. Variousprocess types share similar activity types. Nowtwo more specific order management processtypes need to be implemented. For this purpose,


28 Ulrich Frank

standard order management

prepare contractcontract prepared

order received

calculate terms&conditions term&conditions

satisfactory

no chance for contract

order management for under-specified order

prepare contractcontract prepared

determine optionsrequest received

options determined

calculate terms&conditions term&conditions

satisfactory

no chance for contract

Figure 2: Example of extending a business process type

it would be most helpful to specialise the existingorder management process type.

This would not only allow reusing the respectivemodel and corresponding software implementa-tions, it would also promote safe and convenientmaintenance: Future changes of the core processtype would be immediately effective in the spe-cialised types, too. To satisfy the demand forintegrity, a respective concept of process special-isation would have to satisfy the substitutabil-ity constraint: Every instance of a process typecan act as an instance of the corresponding su-per process type without causing harm (Liskovand Wing 1994). The substitutability constraintis satisfied, if the extensions defined for special-ised concepts are monotonic. This can be accom-plished fairly easy for static abstractions. How-ever, for dynamic abstractions such as businessprocess models adding further activity or eventtypes cannot be monotonic, because it will al-

ways effect the original control flow (Frank 2012).The example in Fig. 2 illustrates this problem.

There are a few approaches in Software Engineer-ing and process modelling which are aimed at arelaxed conception of specialisation of behaviour(Schrefl and Stumptner 2002) or of “workflowinheritance” (Aalst and Basten 2002). Other ap-proaches focus on analysing structural similarit-ies of control flows to promote reuse through pro-cess variants (Koschmider and Oberweis 2007).However, the restrictions these approaches implyremain unsatisfactory (Frank 2012). At the sametime, the still growing relevance of efficientlycreating and maintaining business process mod-els demands for abstractions that allow takingadvantage of similarities.

Request: Future research should aim at conceptsof relaxed process specialisation—which may becombined with instantiation—that promote reusewithout unacceptable restrictions.


While the lack of a sound concept of process spe-cialisation creates a serious problem, the currentstate of business process modelling is even moredreary. The above scenario would suggest to re-use an activity type that was defined already fora certain business process type in a new processtype. However, this is not possible: Every busi-ness process type has to be designed from scratchusing the basic concepts provided by today’s pro-cess modelling languages. Hence, an activitysuch as “prepare contract” cannot be specifiedas a reusable type. Instead it is yet another in-stance of a basic (meta) type like “activity” or“automated activity” that is distinguished fromother primarily through its label. There are ap-proaches that focus on analysing labels in orderto detect similar activity types (Dijkman et al.2011). However, their contribution to reuse islimited: Instead of removing the mess, they trycoping with it.

Request: There is need for extending businessprocess modelling languages and tools with thepossibility to define and reuse activity and eventtypes.

This request is not easy to satisfy. An activitytype is not only defined by its internal structureand behaviour, but also by its context such asthe event that triggers it or the events it pro-duces. Reuse will be possible only, if the contextcan be adapted to some extent. Therefore, therequired concept of an activity type—and of anevent type respectively—must abstract from thecontext without compromising reusability toomuch.

3.2 The Essential Conflict of DesigningDSML

DSML are characterised by convincing advant-ages (Kelly and Tolvanen 2008; Kleppe 2009; Völ-ter 2013). By providing domain-specific concepts,they promote modelling (and programming) pro-ductivity: Modellers are not forced anymore toreconstruct domain-level concepts from genericconcepts such as “entity” or “attribute”. At the

same time, DSML foster model integrity, becausethey prevent the creation of inconsistent mod-els to a certain extent. By featuring a domain-specific concrete syntax, they also promote modelcomprehensibility. Against this background, itdoes not come as a surprise that DSML are re-garded by many as the silver bullet of conceptualmodelling and model-driven software develop-ment. However, their construction is facing adilemma. The more a DSML is tuned to a spe-cific domain, the better is its contribution to pro-ductivity and integrity. However, the more spe-cific a DSML is, the more unlikely it can be usedin a wide range of particular domains. Figure 3illustrates the conflicting effects of semantics onrange of reuse and productivity.

Level of (domain-specific) Semantics

Po

ten

tia

l P

rod

uctivity G

ain

Sca

le o

f R

eu

se

Figure 3: DSML: Illustration of Essential Design Conflict

Some authors suggest to design DSML to theneeds of particular organisations or even pro-jects only (Kelly and Tolvanen 2008; Völter 2013).This recommendation is based on two assump-tions. First, the variety of organisations wouldnot allow for powerful DSML that fit all indi-vidual requirements. Second, there is no needfor further reuse, because creating and using aDSML in one particular project will usually payoff already. Even though both assumptions maybe valid to a certain extent, they are hardly con-vincing. There may be remarkable differencesbetween organisational actions systems and cor-responding terminologies. However, it would bea sign of epistemological defeatism to deny the


30 Ulrich Frank

chance of finding substantial commonalities. Fur-thermore, it can be assumed that actual varietyis also a result of in part arbitrary processes oforganisational evolution, i.e., it is not a reflectionof inevitable differences. Also, there is evidencethat technical languages work in a wide rangeof organisations of a certain kind: The termin-ology used in textbooks will often fit an entireindustry in the sense that it provides a respectedlinguistic structure and serves as as common ref-erence for professionals. Nevertheless, there areorganisation-specific adaptations of textbook ter-minology. They include extensions, refinementsand modifications, some of which may be ques-tionable. The argument that a DSML will alreadypay off in single use scenarios is fine, but it couldstill be much more profitable, since a wider rangeof reuse would allow for much better econom-ies of scale. Therefore, it would be beneficial tocreate hierarchies of DSML, where more specificones are extensions and/or instantiations of moregeneral DSML.

Request: Research on DSML should aim at hier-archies of languages to enable both a wide rangeof reuse and customised languages for narrowdomains.

Figure 4 illustrates the idea of providing mod-elling languages on different classification lay-ers. The highest level (“universal DSML”) corres-ponds to textbook terminology. The concepts onthis level should be applicable to a wide rangeof organisations, hence, promote economies ofscale. The universal DSML should be designedby experts that possess deep knowledge aboutthe general domain as well as rich experiencewith designing DSML. “Local” DSML representmore specific technical languages for organisa-tion modelling that apply to a few organisationsor to one only. They are designed by organisa-tion analysts that are familiar with the respectivedomain. These local DSML that feature a graph-ical notation much like the universal DSML canbe used by authorised managers to specify par-ticular organisational settings. The example alsoshows that there are cases where it makes sense

to create models that include concepts on the M0level.

Universal DSML

M2

Language Designer

Organisational Unit

Position

Specific DSML

(Local „Dialect“)

M1

Organisation Analyst

Department

Team Market Analyst

Particular Organisation

Model

M0

Manager

Marketing DepartmentQuality Circle

Product Group PG 1

Market

Analyst MA2

Market Research

Team

Committee

Quality Circle

Figure 4: Illustration of multi-level modelling languages

Designing such language systems and corres-ponding tools is far from trivial. It requires giv-ing up prevalent architectures of modelling lan-guages that feature a given set of classificationlayers (for respective approaches see Atkinson etal. (2009); Clark et al. (2008); Simonyi et al. (2006).Instead, recursive language models such as the“golden braid” architecture are more promising—and more demanding at the same time, becausethey are not supported by most of today’s devel-opment environments. Apart from that, design-ing languages for enterprise modelling shouldaccount for a further issue. Current DSML areusually specified with metamodels. This is fora good reason: On the one hand, this kind ofspecification corresponds to a paradigm the mod-elling community is familiar with. On the otherhand, it fosters the construction of correspondingtools, because a metamodel can be used as a con-ceptual foundation of a respective modelling tool.The semantics of DSML, e.g., the semantics ofspecialisation concepts, is usually based on the se-mantics of prevalent programming languages tofacilitate the transformation of models into code.However, there are other language paradigmsand specification styles that might enrich enter-prise models. For instance, models designed withlogic-based languages allow for deduction could


enable more sophisticated approaches to analys-ing enterprise models. Since the semantics ofrespective languages, which are typically foundin Artificial Intelligence, is different from thatof DSML used for enterprise modelling today,integrating them into enterprise modelling envir-onments is a demanding task.

4 The Need for Run-Time Use

Originally, enterprise models like most other con-ceptual models were intended for supporting thecreation of information systems only. However,it is obvious that they should be beneficial duringthe entire life cycle of an information system. Ona more generic level, this issue is addressed byresearch on “models at runtime” (Blair et al. 2009).Multi-perspective enterprise models provide ab-stractions of the enterprise that support decisionmaking and other managerial tasks. Also, theycan help people in organisations to develop adeeper understanding of the action system, i.e.,how their work is integrated into a bigger pic-ture. In addition to that, enterprise models canenable users to develop a better understandingof the information system and its interplay withorganisational patterns of action.

4.1 Integrating Enterprise Models withEnterprise Systems

In an ideal case, an enterprise modelling environ-ment would be integrated and synchronised witha corresponding enterprise (software) system.On the one hand, this would enrich enterprisesystems not only with their conceptual founda-tion, but also with a representation of the contextthey are supposed to operate in. On the otherhand, enterprise models would be supplementedwith corresponding instance populations.

This would enable users to navigate from con-cepts on various classification levels to instances—et vice versa. The following scenario illustratesthe benefit of drilling down from an enterprisemodel to instances. A department manager whois new to a firm wants to get a better understand-ing of the way business is done. For this purpose,

he could browse a graphical representation of thecorporate business process map, which shows allbusiness process types, their interrelationshipsand key performance indicators at a glance. Hecould then select a business process type he isinterested in, study the model that describes itsexecution and demand for further aggregate datathat characterises it, such as the number of in-stances per month, average revenues etc. Also, hecould select specific analysis views, e.g., a viewthat associates a selected business process typewith the IT resources it requires. If he was inter-ested in one particular business process type, hecould view the corresponding model. Then, hecould leave the conceptual level and ask for thelist of currently active business processes of thistype and inspect the state of the instances he isinterested in. In addition to that, advanced userscould modify the enterprise system by changingthe enterprise model. The DSML, an enterprisemodel is created with would help preventing ar-bitrary modifications and hence contribute to sys-tem integrity. An outline of a respective system,referred to as “self-referential enterprise system”is presented in (Frank and Strecker 2009). Figure 5illustrates the idea of integrating enterprise sys-tems with enterprise modelling environments.

Such a system would allow realising the vision ofinteractive models propagated by Krogstie (2007,p. 306): “The use of interactive models is aboutdiscovering, externalising, capturing, expressing,representing, sharing and managing enterpriseknowledge.” In other words: It would be a con-tribution to empowering people who work inand interact with organisations. The realisationof self-referential enterprise systems does notonly require developing further DSML, but alsoredesigning enterprise software systems.

Request: To further exploit the potential of bothenterprise software systems and enterprise mod-elling environments, research should aim at de-veloping the foundations for integrating bothkind of systems into a versatile tool for managingand adapting an organisation and its informationsystem.


32 Ulrich Frank

Is there a chance for outsourcing services?

Is there a chance for outsourcing services?

M2

M1

M0

Figure 5: Navigating an enterprise model and corresponding instances

4.2 Deficiencies of PrevalentProgramming Languages

The integration of enterprise modelling envir-onments and enterprise systems does not onlyrequire research on enterprise models and theirrepresentation. It also demands for system ar-chitectures that cannot be satisfactorily accom-plished with prevalent programming languages.Integration implies common representations ofshared concepts. In today’s modelling environ-ments, conceptual models are usually represen-ted by objects on the M0 level—even though theybelong to the M1 or even a higher level. Over-loading the M0 level happens for a good reason:Prevalent programming languages are restrictedto the dichotomy of objects and classes. Hence,there are no meta classes that were required tospecify classes—and that would allow treatingclasses as objects, too. Therefore, a commonrepresentation of classes in both systems is notpossible. Instead, the only way to associate amodelling environment with a corresponding en-terprise system would be to generate code, i.e.,

classes in the enterprise system, from objectsin the enterprise modelling environment. As aconsequence, one would have to deal with thenotorious problem of synchronising models andcode. Figure 6 illustrates how the M0 layer ofmodelling tools is overloaded and that conceptsin modelling tools are located on a classificationlayer that is different from that of correspondingconcepts in an associated enterprise informationsystem.

Recent developments in research on program-ming language has produced (meta) program-ming languages that were designed for creatingdomain-specific programming languages. Lan-guages like XMF (Clark and Willans 2012; Clarket al. 2008) are especially promising, since theyallow for an arbitrary number of classificationlevels, which enables a common representationof models and respective code. Hence, modify-ing an enterprise model implies modifying therespective part of the enterprise software simul-taneously.


Metamodel Editor

concepts of metamodelling language

Implementation Layers

Enterprise Information System

Enterprise Modeling Environment

corresponds to

concepts of metamodel M2M0

M3M1

Model Editor(s)

concepts of modelling language

concepts of model M1M0

M2M1

schema (classes, types)

runtime (objects, data)M0

M1

name: String

...

MetaEntity

name: String

averageSalary: Money

availability: Level

...

Position

Programmer

average: € 48.300

availability: #low

id: String

staffed: Boolean

...

Programmername: String


availability: Level

...

Position

name: String


availability: Level

...

Position

staffed = true

...

p1: Programmername = 'Programmer'

averageSalary = 48.300

availability = #low

...

pos3: Position

Figure 6: Mismatch of Classification Levels

Request: To advance the state of current model-ling environments, research needs to focus ontools that overcome the limitations of currentprogramming languages.

5 The Need for Collaboration

Extending the scope of enterprise models anddeveloping them to an omnipresent represent-ation of organisations requires an amount ofresearch and development that cannot be car-ried out by the current enterprise modelling com-munity alone. To advance the field, there is needfor cross-disciplinary collaboration and for accu-mulating resources.

5.1 The Importance of BundlingResources

The development of comprehensive enterprisemodels will overburden most organisations. Thisis the case, too, for respective DSML. Referenceenterprise models and wide-spread DSML aresuited to effectively address this problem. Fur-thermore they would provide a foundation forcross-organisational integration of action sys-tems and information systems, which could en-able a tremendous boost of productivity. At thesame time, the development of reference artefactsthat combine a descriptive and a prescriptive ap-proach constitute an attractive research goal: It is

characteristic for research to abstract from singlecases and aim at constructions that work for anentire class of cases. Furthermore, applied re-search is motivated by improving existing prac-tice with respect to certain goals. Unfortunately,the development of reference enterprise modelsand corresponding languages and tools requiresresources that are not available to a single re-search institute. Furthermore, establishing anddisseminating them in practice depends on eco-nomic and political aspects that are beyond theabilities and intentions of academics. Against thebackground, it is obvious that there is need tobundle resources of research institutions. At thesame time, it is necessary to get vendors of en-terprise software and prospective users involved.On the one hand, they need to be involved to sup-port requirements analysis. On the other hand,using reference artefacts in practice is the onlyway to promote their dissemination. Unfortu-nately, there are serious obstacles that impedeboth bundling of research resources and involve-ment of companies. Research is based on compet-ition and the idea of scientific progress. Collab-oration of research institutes implies to give upcompetition to a large extent. At the same time,for reference artefacts to be beneficial they needto be consolidated—which may jeopardise sci-entific progress. While there are probably manyvendors and client organisations that would be


34 Ulrich Frank

happy to use reference models, most of them willlikely be reluctant to participate in respectivedevelopment projects, since the return on suchan investment would be hard to determine. Nev-ertheless, to promote the benefit of enterprisemodels, reference artefacts that enable attractiveeconomies of scale are of pivotal relevance.

Request: There is need for initiatives to collabor-atively develop and disseminate reference arte-facts. They need to provide convincing incentivesboth for academics and practitioners.

One of the prime examples of community-drivencollaboration is free and open source software(FOSS). Respective initiatives have successfullypromoted collaboration of developers and users.Also, they led to software systems of surprisingquality, and, in some cases, to an impressive dis-semination. Inspired by the apparent success ofsome FOSS projects, corresponding “Open Mod-els” initiatives have been proposed (Frank andStrecker 2007) and inspired the creation of openmodel repositories (France et al. 2007), (www.openmodels.org, www.openmodels.at). Whilethese repositories have triggered remarkable at-tention, there is still need for more active parti-cipation.

5.2 Enterprise Models as Object andPromoter of Cross-DisciplinaryCollaboration

Enterprise models are aimed at providing a me-dium to foster communication between stake-holders with different professional background.On the one hand that requires reconstructingtechnical languages and professional patterns ofproblem solving. On the other hand it recom-mends analysing how prospective users reactupon the models they are presented with. Thatincludes concepts as well as their designationand (graphical) representation. For using enter-prise models effectively, software tools are man-datory. Developing and integrating them withother enterprise software systems creates sub-stantial challenges for Software Engineering or—in other words—interesting research questions.

Enterprise modelling is not an end itself. Instead,it is supposed to have a positive impact on a com-pany’s economics and competitiveness. However,assessing the costs of creating and maintainingenterprise models, which may include the devel-opment of languages and tools, is not a trivialtask—and it is similarly challenging to determ-ine the benefits that can be contributed to thedeployment of enterprise models. Apart fromeconomic effects, the extensive use of enterprisemodels within an organisation may have an im-pact on how people perceive not only their tasksbut also the entire organisation. The increasein transparency may have an effect on estab-lished patterns of organisational power and mayrequire new approaches to managing organisa-tions. Against this background it is obvious thatenterprise modelling involves a wide range ofdemanding research questions that concern vari-ous disciplines. Business and Administration ingeneral is aimed at developing and improving ter-minologies and methods that are suited to struc-ture and guide purposeful action in enterprises.Various subfields, such as Financial Management,Accounting, Industrial Management, Logisticscould contribute to extend and deepen enterprisemodels. In Psychology, the interaction betweencognitive models and external representationsis a core research topic. Applied to enterprisemodelling this would include the question howconceptual models effect individual and collect-ive decision making. That includes analysing theimpact of graphical notations on people’s abilityto understand complex matters—and the develop-ment of guidelines for designing notations thatfit certain cognitive styles. Both, from a psycho-logical and a sociological point of view, it wouldbe interesting to analyse how enterprise modelseffect the social construction of reality, i.e., towhat extent people perceive the model as the en-terprise and what that means for the way they(inter) act. Assuming that enterprise models mayhave a substantial effect on an organisation’s per-formance implies challenging research questionsfor economic studies that are not restricted toenterprise models, but comprise the economics

www.openmodels.org

www.openmodels.org

www.openmodels.at


of models and methods in general. Combining re-search results from various disciplines would notonly contribute to advance our knowledge aboutenterprise models and our ability to use them ef-fectively, it is also suited to enrich the state of theart in the participating disciplines, since it wouldintegrate it with contributions from other fields.Therefore the following request could have a bet-ter chance to succeed than yet another call forinter-disciplinary research.

Request: Advancing the field of enterprise model-ling recommends to establish inter-disciplinaryresearch collaboration.

6 Conclusion

In the past, enterprise modelling, though argu-ably pointing at a core topic of Information Sys-tems, has been subject of a rather small, special-ised research community. In Business and Ad-ministration it is regarded as too much focussedon technical aspects by some, while some tradi-tionalist colleagues in Computer Science suspectit of lacking formal rigour. However, enterprisemodelling is more than analysing and designinginformation systems—and it is certainly muchmore than drawing “‘bubbles and arrows”. Enter-prise modelling is about conceptualising an im-portant part of the world—as it actually is and asit might be. Hence, it requires knowledge abouthow people (inter) act in organisations, how in-formation systems infrastructures are built—andthe creativity to develop substantial images ofattractive future worlds that comprise the pur-poseful construction and use of information sys-tems. It is about how we perceive the worldwe live and work in and how we think about itand might change it—alone and together withothers. In addition to supporting collaborationbetween stakeholders with different professionalbackgrounds in organisations, enterprise modelsmay also serve as a medium and object of inter-disciplinary research. At the same time, theyare suited to foster the exchange between prac-tice and academia, because they allow to integ-rate more abstract representations of enterprises

with more specific ones. Last but not least, enter-prise models provide a laboratory for learning,because they convey a solid conceptual found-ation of information systems and surroundingaction systems—and enable students to navigatethrough an enterprise on different levels of de-tail and abstraction. With respect to such a wideand deep conception of enterprise modelling it isimportant not only to identify relevant steps offuture research, but also to spread the word andencourage others to participate in joint projects.Further developing the field also requires to putmore emphasis on assessing model artefacts. Onthe one hand that comprises the development ofpragmatic criteria to evaluate models and model-ling languages with respect to an intended prac-tical use. On the other hand, it relates to assess-ing the epistemological quality of model artefactsas research results. Developing and applying re-spective criteria is an important prerequisite ofscientific competition and progress.

In this paper I gave a personal account of thetopics we should address in the next years toadvance our field. It is needless to say that otherrelevant topics exist, too. I would hope that therequests presented in this paper contribute to adiscourse on our future research agenda.

References

der Aalst W., Basten T. (2002) Inheritance ofWorkflows – An Approach to Tackling Prob-lems Related to Change. In: Theoretical Com-puter Science 270, p. 2002

Atkinson C., Gutheil M., Kennel B. (2009) A Flex-ible Infrastructure for Multilevel LanguageEngineering. In: IEEE Transactions on Soft-ware Engineering 35(6), pp. 742–755

Blair G., Bencomo N., France R. B. (2009)Models@ run.time: Computer. In: Computer42(10), pp. 22–27

Brinkkemper S. (1996) Method Engineering: En-gineering of Information Systems Develop-ment Methods and Tools. In: Information andSoftware Technology 38(4), pp. 275–280


36 Ulrich Frank

Buckl S., Matthes F., Roth S., Schulz C., SchwedaC. (2010) A Conceptual Framework for Enter-prise Architecture Design. In: Proper E., Lank-horst M. M., Schönherr M., Barjis J., Over-beek S. (eds.) Trends in Enterprise Architec-ture Research. Lecture Notes in Business In-formation Processing Vol. 70. Springer, Ber-lin, Heidelberg and New York, pp. 44–56

CIMOSA: Open system architecture for CIM.Springer, Berlin, Heidelberg and New York

Clark T., Willans J. (2012) Software LanguageEngineering with XMF and XModeler. In:Mernik M. (ed.) Formal and Practical Aspectsof Domain-Specific Languages. InformationScience Reference, pp. 311–340

Clark T., Sammut P., Willans J. (2008) Superlan-guages: Developing Languages and Applica-tions with XMF. Ceteva

Dijkman R., Dumas M., van Dongen B., KäärikR., Mendling J. (2011) Similarity of businessprocess models: Metrics and evaluation. In:Inf. Syst. 36(2), pp. 498–516

Ferstl O. K., Sinz E. J. (1998) Modeling ofbusiness systems using the Semantic ObjectModel (SOM): A methodological framework.In: Bernus P., Mertins K., Schmidt G. (eds.)Handbook on Architectures of InformationSystems. International Handbooks on Inform-ation Systems Vol. 1. Springer, Berlin, Heidel-berg and New York, pp. 339–358

Fettke P., Loos P. (eds.) Reference Modeling forBusiness Systems Analysis. Idea Group, Her-shey

France R., Bieman J., Cheng B. (2007) Repositoryfor Model Driven Development (ReMoDD).In: Kühne T. (ed.) Models in Software En-gineering. Lecture Notes in Computer Sci-ence Vol. 4364. Springer Berlin Heidelberg,pp. 311–317

Frank U. (1994) Multiperspektivische Un-ternehmensmodellierung: Theoretischer Hin-tergrund und Entwurf einer objektorientier-ten Entwicklungsumgebung. Oldenbourg,München

Frank U. (2011) The MEMO Meta ModellingLanguage (MML) and Language Architecture.

2nd Edition. ICB-Research Report 43Frank U. (2012) Specialisation in Business Pro-

cess Modelling: Motivation, Approaches andLimitations. 51. University of Duisburg-Essen.Essen

Frank U. (2013a) Domain-Specific Modeling Lan-guages - Requirements Analysis and DesignGuidelines. In: Iris Reinhartz-Berger, AronSturm, Tony Clark, Yair Wand, Sholom Co-hen, Jorn Bettin (eds.) Domain Engineering:Product Lines, Conceptual Models, and Lan-guages. Springer, pp. 133–157

Frank U. (2013b) Multi-Perspective EnterpriseModeling: Foundational Concepts, Prospectsand Future Research Challenges. In: Softwareand Systems Modeling (in print)

Frank U., Strecker S. (2007) Open Reference Mod-els – Community-driven Collaboration toPromote Development and Dissemination ofReference Models. In: Enterprise Modellingand Information Systems Architectures 2(2),pp. 32–41

Frank U., Strecker S. (2009) Beyond ERP Systems:An Outline of Self-Referential Enterprise Sys-tems: Requirements, Conceptual Foundationand Design Options

Group T. O. (2009) TOGAF Version 9. Van Haren,Zaltbommel

Henderson-Sellers B., Ralyté J. (2010) SituationalMethod Engineering: State-of-the-Art Re-view. In: Journal of Universal Computer Sci-ence 16(3), pp. 424–478

Kelly S., Tolvanen J.-P. (2008) Domain-specificmodeling: Enabling full code generation.Wiley-Interscience and IEEE Computer Soci-ety, Hoboken and N.J

Kleppe A. G. (2009) Software language engin-eering: Creating domain-specific languagesusing metamodels. Addison-Wesley, UpperSaddle River and NJ

Koschmider A., Oberweis A. (2007) How to de-tect semantic business process model vari-ants? In: Proceedings of the 2007 ACM sym-posium on Applied computing. SAC ’07.ACM, New York, NY and USA, pp. 1263–1264

Krogstie J. (2007) Modelling of the People, by the


People, for the People. In: Krogstie J., OpdahlA., Brinkkemper S. (eds.) Conceptual Mod-elling in Information Systems Engineering.Springer Berlin Heidelberg, pp. 305–318

Land M. O., Waage M., Proper E., Cloo J.,Steghuis C. (2009) Enterprise architecture:Creating value by informed governance.Springer, Berlin

Lankhorst M. M. (2005) Enterprise architectureat work: Modelling, communication, and ana-lysis. Springer, Berlin, Heidelberg and NewYork

Liskov B. H., Wing J. M. (1994) A BehavioralNotion of Subtyping. In: ACM Transactionson Programming Languages and Systems 16,pp. 1811–1841

Moody D. L. (2009) The “Physics” of Notations:Toward a Scientific Basis for ConstructingVisual Notations in Software Engineering. In:IEEE Transactions on Software Engineering35(6), pp. 756–779

Ralyté J., Agerfalk P. J., Kraiem N. (2005)Methods, Techniques and Tools to SupportSituation-Specific Requirements EngineeringProcesses: SREP’05. University of Limerick,Limerick

Ralyté J., Brinkkemper S., Henderson-Sellers B.(2007) Situational method engineering: Fun-damentals and experiences. Proceedings ofthe IFIP WG 8.1 Working Conference, 12-14September 2007, Geneva, Switzerland. IFIP -the International Federation for InformationProcessing Vol. 244. Springer, New York

Scheer A.-W. (1992) Architecture of IntegratedInformation Systems: Foundations of Enter-prise Modelling. Springer, Berlin and NewYork

Schrefl M., Stumptner M. (2002) Behavior-consistent specialization of object life cycles.In: ACM Transactions on Software Engineer-ing Methodologies 11(1), pp. 92–148

Simonyi C., Christerson M., Clifford S. (2006)Intentional software. In: Proceedings of the21th Annual ACM SIGPLAN Conference onObject-Oriented Programming, Systems, Lan-guages, and Applications (OOPSLA 2006).

ACM, pp. 451–464Strecker S., Frank U., Heise D., Kattenstroth H.(2012) MetricM: a modeling method in sup-port of the reflective design and use of per-formance measurement systems. In: Inform-ation Systems and e-Business Management10(2), pp. 241–276

Völter M. (2013) DSL Engineering: Designing,Implementing and Using Domain-SpecificLanguages. dslbooks.org

Zachman J. A. (1987) A framework for inform-ation systems architecture. In: IBM SystemsJournal 26(3), pp. 276–292

Ulrich Frank

Chair of Information Systems and EnterpriseModellingInstitute for Computer Science and BusinessInformation SystemsUniversity of Duisburg–EssenUniversitätsstraße 945141 [email protected]


38 José Tribolet and Pedro Sousa and Artur Caetano

José Tribolet and Pedro Sousa and Artur Caetano

The Role of Enterprise Governance and Cartographyin Enterprise Engineering

Enterprise artography is fundamental to govern the transformation processes of an organisation. The artefactsof enterprise cartography represent the structure and dynamics of an organisation from three temporal views:as-was (past), as- is (present), and to-be (future). These views are dynamically generated from a continuousprocess that collects operational data from an organisation. This paper defines a set of enterprise cartographyprinciples and provides an account of its role in understanding the dynamics of an organisation. The principlesare grounded on control theory and are defined as a realisation of the observer and modeller componentsof the feedback control loop found on dynamic systems. As a result, an organisation can be abstracted asa dynamic system where a network of actors collaborate and produce results that can be depicted usingcartographic maps.

1 Introduction

This paper explores the role played by enterprisecartography and enterprise governance withinthe enterprise engineering discipline. Enterprisegovernance relates to enterprise transformationsince the change of operational processes, re-sources and business rules define new manage-ment boundaries (Hoogervorst 2009). Enterprisearchitecture contributes to enterprise transform-ation as it enables modelling the organisation’sstructure and dynamics along with the underly-ing restrictions and design principles (Lankhorst2013; Op’t Land 2009). Transformation is oftenseen as the set of initiatives that change the or-ganisation’s domain from the current as-is stateto an intended to-be state. These two states de-scribe organisational variables at different mo-ments in time. The as-is state is defined by thevariables that changed due to past events, whilethe to-be state specifies an expected state config-uration of the organisational variables. Betweenthese two events, the organisation reacts to otherevents that are triggered by the operation of thetransformation processes.

The issues we address in this position paper focusin the ability to observe and govern the organ-

isation during such transition. This is importantbecause during each transformation initiative anorganisation has to react to events. Some of theseevents may be unrelated to the transformationinitiative but may impact the transformation pro-cess and therefore deviate the organisation fromachieving the planned future state.

This paper presents two contributions. The firstis defining enterprise cartography as a functionof the observer and modeller roles as defined bythe enterprise’s dynamic feedback control loop.Enterprise cartography is not associated withthe enterprise design, but with the abstractionand representation of the enterprise reality. Al-though this differentiates enterprise cartographyfrom enterprise architecture, it may be correctlypointed out that cartography is part of enterprisearchitecture. But given the relevance of carto-graphy to understand the dynamics of the feed-back control loop of an organisation, we opted todiscuss the concerns of cartography separatelyfrom those of enterprise architecture. The secondcontribution of the paper is stating the empiricalprinciples that ground the design of the carto-graphy process to play the role of the observerand modeller in the enterprise dynamic feedback

Enterprise Modelling and Information Systems ArchitecturesVol. 9, No. 1, June 2014The Role of Enterprise Governance and Cartography in Enterprise Engineering 39

control loop. Dynamic systems and enterprisegovernance are described in Sect. 2 and Sect. 3.Section 4 presents enterprise cartography.

2 Dynamic Systems

The application of systems theory to systemsengineering has been discussed since the 1970s(Eriksson 1997; Moigne 1977). Systems theoryrelates to organisational systems mainly throughthe principles of dynamic systems, especially con-trol feedback loops (Abraham et al. 2013; San-tos et al. 2008). These concepts can be furthercombined with classic management theories asa means to clarify how feedback loops interactwith different organisational views, such as gov-ernance, management, and operations (Fig. 1).

In control theory, the modeller presents a systemview that specifies its current as-is state (Levine1996). The current state makes possible to estim-ate a future state of the system in the absence ofunexpected events. To handle the potential devi-ations that occur from such events, control the-ory introduces the concept of controller. The con-troller analyses the continuous stream of eventsand modifies the system’s controllable variablesas a means of keeping the system behaving asplanned (Fig. 2). This is similar to the control of aphysical body moving toward a target: the mod-eller determines the current position and speedof the object and feeds it to the controller; if anunexpected event occurs, then the controller cor-rects the movement of the object by applying thenecessary forces and thereby ensuring that thetarget is reached.

We argue that the relationships between enter-prise governance and enterprise cartography canbe established using the principles of dynamicsystems feedback control, where cartographyplays the aforementioned roles of observer andmodeller. These relationships are explained inthe next two sections.

3 Enterprise Governance

An enterprise is a network of independent act-ors. Actors collaborate with other actors alongtime and thus create a dynamic collaborative net-work. Actors also produce autonomous beha-viour that may change the overall state of thesystem. Actors can be classified ascarbon-basedactors, i.e. humans, and silicon-based actors, i.e.computers. This network runs within a domainwhere the independent actors behave towards afuture state of affairs, and thus produce events,some of which may be unexpected. Therefore,all enterprise domain state changes are a con-sequence of the individual behaviour of an actoror of the composite behaviour that derives fromthe actor collaborations. These collaborationsmay occur between actors that are enclosed bythe organisation’s boundary, or between an actorthat is external to the organisation and one in-ternal actor. So, the behaviour of an enterprise“is” a result of what “it does”. An enterprise cantherefore be regarded as a large “bionic” distrib-uted network of carbon-based and silicon-basedactors that are continuously interacting and pro-ducing behaviour.

The current technological advances make pos-sible near real-time, transparent and ubiquitousinteraction between people and systems. As such,the boundary between manual, semi-automatedand even some automated operations becomesblurred. This means that the actions performedby people cannot be easily separated from thoseof people supported by a network of computers,and from those of networks of computers. Thesecollaborations can be abstracted as the result of asingle network that operates in (near) real-time.The actors that interact within this network actautonomously.

Autonomous behaviour is evident from how aperson acts within an organisation since the statechange produced by a human actor can only beobserved after the action is concluded. But thesame phenomenon is also observed on informa-tion systems because one can only assert what



Figure 1: Organisational views and feedback loop, adapted from (Abraham et al. 2013).

Figure 2: A single-input, single-output feedback loop.

a computer actor has produced after the actualaction is performed. The degree of predictabil-ity of automated computer actions is potentiallyhigher than that of humans. But achieving cer-tainty is not feasible due to a number of factors.On the one hand, a system may not behave asexpected due to faults or failures. And even inthe absence of faults of failures, the system maybe misaligned with the business. On the otherhand, the interaction between multiple systemscan produce emergent behaviour, meaning thatthe overall behaviour of the system may not bethe linear sum of each individual behaviour unit.As a result, there is a potential gap between theresults that derive from planned actions and theactual results. This makes it impossible to fullyestimate the outcome of the interactions in sucha network.

This reasoning supports the conclusion that en-terprises are dynamic systems. Enterprises areactually a system of systems, composed of andpart of other dynamic systems. As such, thereis an opportunity to try to understand an enter-prise as a complex system through the lenses ofsystems theory, in particular through the bodyof knowledge of systems theory and dynamicsystems control. However, this application mustalways consider the intrinsic bionic nature of anenterprise, as people cannot be dissociated fromits essence. We defend that all this body of know-ledge is directly applicable to enterprises throughenterprise engineering. The fundamental pur-pose of engineering is to provide humans withartefacts that augment their individual and col-lective capability to deal with specific situations.Engineering helps humans to understand realityand to pro-actively and purposefully transformit as idealised by individual and collective goals.This is the primary purpose of enterprise engin-eering (Dietz et al. 2013).

How do systems theory and dynamic systemscontrol relate to enterprise engineering? Well,let us start with the “bionic state machine” meta-


phor presented earlier. According to systemstheory, this model can be abstracted as two sep-arate subsystems: a feed-forward action system,which is combinatorial in nature and transformsinputs into outputs, and a feed-back cyberneticsystem, which uses as input the state observa-tions and results provided by the feed-forwardaction system. The feed-back system uses thisinformation to continuous estimate the currentstate of the system. This is accomplished by con-textualising the observations, i.e. by situating theobservations into the semantic model of the sys-tem. Based on these observations, the feed-backcybernetic system then decides on the actionsthat all the actors of the system must performin order to keep the system on a trajectory thatachieves its goals. This process is continuouslyperformed. These concepts have been extens-ively applied to most engineering areas for atleast half a century (Andrei 2005).

In this paper we hypothesise that the applicationof control theory is useful to help understandingenterprise engineering. The next hypotheticalprinciples characterise enterprise governance asa dynamic systems theory problem.

Principle 1 Actions performed by people areenacted by the feed-forward action system.People play multiple actor roles within anenterprise such as operational, middle man-agement, knowledge work, auditing, advisory,governance or executive roles. If an enter-prise is abstracted as a layered system, allthese actions occur at the operational layer,where actual operations are performed by act-ors. People are abstracted as actors playingroles within well specified semantic domainsthat uniquely define their contexts of indi-vidual action and interaction (Caetano et al.2009; Zacarias et al. 2005). An actor is capableof playing several roles simultaneously.

Principle 2 A person can be abstracted as a sys-tem of systems whenever its actions and in-teractions occur within the enterprise net-work. This means that the roles played bypeople are subject to the rules of the dynamic

systems control model. The actions of a per-son are the result of a combinatorial proced-ure: a person observes the world, attemptsto contextualise and understand its meaning,and then performs an action. This procedurecorresponds to the role of controller. By act-ing as a controller, the person can correct thedeviations between the current state and theintended state. As such, to achieve goals anhuman actor operates his own local feed-backsubsystem. These actions do not occur at theoperational layer but at an higher layer thatplans and controls the operations (Abrahamet al. 2013).

Principle 3 An enterprise is more than the sumof its actors and resources. Organisationalfactors such as culture, values, power, and hier-archical structures are elements in defining anenterprise. We abstract these “soft” factors asquality requirements that constrain and para-metrise the operating system of an humanactor. They are key determinants to the waya human interprets the observations of reality,as well as he reads these observations throughhis own models of the world, based on whichhis own sense making operates. These factorshave impact on the actions of a human actorsince they change how it plays the controllerrole.

Principle 4 Enterprise self-awareness requiresthe specification of the domain of action. Thisis the realms of enterprise governance. Gov-ernance actions are distinct from executive,managerial, and operational actions, becausethey are geared towards the preservation ofthe enterprise self-awareness. Hence, gov-ernance focuses on the design rules and prin-ciples that constrain the enterprise actors, alongwith their actions and interactions.

Principle 5 Maintaining the enterprise as a singleentity requires actors to dynamically main-tain a view of the actual state of the enter-prise.

The previous principles state the relationshipsbetween an individual actor and its own dynamic



control system. But how do the multiple actors,either carbon or silicon, interact and producecomposite behaviour? Using a metaphor: whatmakes a group of heterogeneous and autonomousmusicians become a musical ensemble? Why isthis collective entity more than the linear sum ofits individual parts? So, what defines the bound-ary of an enterprise? What forces bind togetherits autonomous actors as a single entity? Webelieve that the answer to this question lies inthe enterprise’s “semantic model of itself”. Wecall this enterprise self-awareness (Abraham etal. 2013; Potgieter and Bishop 2003; Santos et al.2008). This means that if an enterprise has a com-mon semantic model of its actors then in becomesa single collective entity. If there is no commonsemantic model then the actors are unable tobe self-aware of their context and as a result nosingle collective entity can be defined. This se-mantic model is a shared dynamic model that isconstantly updated by all its active components.It is precisely this shared semantic model thatdefines a musical ensemble: each musician hasits own role, but both individually and as a wholethey are self-aware that they share the goal ofplaying the same piece of music according to aset of rules.

The systemic nature of an enterprise and its cy-bernetic attributes stress the need for having en-gineering artefacts to support the collective un-derstanding of its changing reality. Enterprisecartography is fundamental to support this task.Furthermore, this enterprise engineered augmen-ted capability is essential to support the increas-ing challenges of enterprise governance, whichare essential to preserve the integrity of an en-terprise as a collective entity. The next sectiondescribes the goals of principles of enterprisecartography.

4 Enterprise Cartography

Cartography is the practice of designing andcreating maps. It is based on the premise that

reality can be modelled in ways that commu-nicate information effectively. Enterprise carto-graphy deals with providing up-to-date model-based views of an enterprise architecture and itsgoal is facilitating its communication and ana-lysis. We have been successfully applying enter-prise cartography concepts to enterprise architec-ture projects (Caetano and Tribolet 2006; Caetanoet al. 2009, 2012b; Sousa et al. 2007, 2009) and de-veloping computer-based tools to support enter-prise cartography (Caetano et al. 2012b; Filipeet al. 2011; Sousa et al. 2011). Currently, the prin-ciples described here are implemented in a com-mercial tool that is being used in several mediumand large scale enterprise architecture projects1.This section describes some empirical findingsthat we have observed in these cases.

The concept of abstracting reality through repres-entations is not limited to engineering disciplines.Cartography itself is an established disciplinethat has played a major role in the developmentof mankind. Cartography is an abstraction pro-cess that systematically and consistently trans-forms an observation of reality into a map or agraphical representation. The production of amap embraces many different concerns, includ-ing scientific, technical, and purely aesthetic. En-terprise cartography denotes the discipline thatdeals with the conception, production, dissemin-ation and study of the maps of an enterprise tosupport its analysis and collective understanding.

Classic cartography is usually associated withthe representation of static objects, as in the caseof geographic maps. Modern cartography dealswith the representation of both static and dy-namic objects and is commonly grounded in in-formation science, geographic information sci-ence and geographic information systems. Car-tography must also provide multiple consistentviews of the same system. For example, geo-graphical maps often combine different views,such as political boundaries, topographic featuresand several other features. This entails defining

1http://www.link.pt/eams/

http://www.link.pt/eams/


Figure 3: Relationships between meta-model, views, viewpoints, diagrams, and stakeholders, adapted from (The OpenGroup 2009).

abstraction rules and classification mechanismsso that all of views are consistent. The carto-graphy of dynamic objects also requires to ab-stract the rules that constrain how objects changeand relate to each other over time.

Enterprise cartography deals with the dynamicdesign and production of architectural views thatdepict the components of an organisation andtheir dependencies. It shares its constructs withenterprise architecture, such as meta-models, mod-els, views, repositories, frameworks, and designrules. However, its goal is descriptive. A view ex-presses the architecture of a system from the per-spective of concerns defined by its stakeholders.Views are defined by viewpoints, which establishthe conventions for the construction, interpreta-tion and use of architecture views (ISO/IEC/IEEE2011; The Open Group 2009). Figure 3), takenfrom the TOGAF 9 specification, illustrates thebasic relationships between these concepts. Thefollowing principles distinguish cartography fromenterprise architecture.

Principle 1 Enterprise cartography uses obser-vations to produce the representations of an

organisation. The process of organisational

data collection is a core concern of enterprise

cartography. Data collection is not a concern

of the mainstream approaches to enterprise

architecture.

Principle 2 Enterprise cartography focus on thedynamic description of an organisation. It

does not deal with the processes or governance

of organisational transformation. The pur-

poseful transformation of organisations is ad-

dressed by enterprise architecture.

Principle 3 Enterprise cartography keeps up-to-date architectural views. This implies auto-

mated or supervised data collection and view

creation. Ideally, these tasks should be per-

formed at the same frequency as that of or-

ganisational change. Enterprise architecture

techniques do not aim to provide systematic

support for data collection nor the automated

design and creation of views, meaning these

tasks are usually manual and creative.



4.1 Approaches to EnterpriseCartography

There are several approaches to generate organ-isational models from the data extracted fromenterprise systems. Configuration ManagementDatabases (CMDB), as defined by ITIL (Adams2009), manage the configurations and relation-ships of information systems and technologicalinfrastructure. To populate a CMDB, some solu-tions provide auto-discovery techniques that de-tect nodes, virtual machines and network devicesto create infrastructural views. Auto-discoveryis actually a cartographic process and requiresthat the type of the concepts to be discoveredis specified in advance (Filipe et al. 2011). Theresulting CMDB instance will contain a partialmodel of the organisation’s infra-structure. Thismodel can be communicated through differentbut consistent visualisation mechanisms, suchas textual reports or graphical models that aredesigned according to a symbolic notation anddesign rules (Lankhorst 2013).

At the business and organisational layer thereare several cartographic techniques defined bybusiness process management (Dumas et al. 2013)and process mining (Aalst et al. 2012). These tech-niques make use of event logs to discover processactivities, control and data flows, as well as or-ganisational structures (Aalst 2011; Aalst et al.2012; Agrawal et al. 1998). In this case, discoveredprocesses correspond to actual instances of pro-cesses, not to the designed processes. Modelanalysis can also be used to assess the conform-ance of processes against constraints (Caetano etal. 2012a; Molka et al. 2014). Another example ofenterprise cartography is the inference of inter-organisational processes based on EDI event logs(Engel et al. 2012). Semantic technologies, suchas ontologies, can also be used to analyse enter-prise models (Antunes et al. 2013, 2014). Businessintelligence techniques that collect data from or-ganisation systems to produce reports and dash-boards are another example of cartography (Neg-ash 2004). Business intelligence actually supports

the feed-back control loop by providing man-agers with a model of the organisation that al-lows them to ground their actions and decisions.

Enterprise cartography is already a reality in sev-eral domains. However, handling dynamic ob-jects, time and change is not explicitly addressedby most approaches. We aim at a generic andsystemic approach, very much in line with theconcept of ”Enterprise Architecture Dashboard”(Op’t Land 2009), that displays the enterprise cur-rent and future states, its performance and thedirections of the organisation transformation pro-cess.

4.2 Principles of EnterpriseCartography

This section describes a set of principles thatdefine Enterprise Cartography. These principlesuse the following definitions.

Project is an transformation process designed toachieve a goal specified by a to-be state.

Organisation variable references specific inform-ation or a value associated to an organisationalartefact.

Organisation state contains the values of a sub-set of organisation variables at a given point intime.

As-was state is the set of all organisation statesobserved in a specific point in the past.

As-is state is the set of organisation states asobserved in the current point in time.

To-be state is the set of organisation states thatare predicted to occur in a specific point in thefuture.

Principle 4 The as-is state is defined by the as-was and to-be states.Memory of the past state (as-was) and the fu-ture state (to-be) define the behaviour of an or-ganisation. The to-be state specifies the goalsof transformation projects. Without the to-bestate the transformation processes cannot beexecuted or measured since no project goalsare defined.


Principle 5 The definition of the to-be state al-ways precedes the definition of the as-isstate.Organisational artefacts must be always de-fined as goals in the to-be state before beingcaptured in the as-is state. This means thatthe organisational artefacts are not createdincidentally but always as the result of a trans-formation project.

Principle 6 All organisational artefacts can beclassified as being in one of four invariantstates.Gestating is the state that describes an organ-isation artefact after it is conceived, i.e. afterit starts being planned, designed or produced.At this state, the artefact does not yet exist asan active element of the organisation in thesense it is not yet able to produce behaviourbut can be passively used by organisationaltransactions and processes.Alive is the state that an artefact enters afterbirth. Birth is the event that signals the mo-ment when a gestating artefact enters the alivestate. This means that the artefact is now ableto produce behaviour as part of the organisa-tional transactions and processes.Dead is when a gestating or alive artefact is in-active in the sense it is no longer able to play arole in the organisational transactions and pro-cesses. This state is the opposite of gestationthat brought the artefact into existence. How-ever, a dead artefact may still have impact onthe organisation. For example, an applicationor server enter the dead state when they stopoperating and will remain in that state untilthey are fully retired from the organisationalinfrastructure.Retired represents the post-death state wherethe artefact is unable to further interact withother artefacts.Organisational artefacts exist first in the to-bestate and only then in the as-is state. This ap-plies to each state transition of the artefact’slife-cycle. Artefacts are conceived as the futureresult of a project, thereby entering the gest-

ating state. They remain in this state until theproject successfully completes. After that theartefact becomes alive. An alive artefact dieswhen a decommissioning project completes. Agestating artefact can also die if the project iscancelled or not completed. A dead artefactis retired when a retirement project explicitlyremoves the artefact from the organisationalstructure. Therefore, all state changes apply-ing to an artefact are the result of a transform-ation project. As such, the to-be state alwaysprecedes the as-is state (Sousa et al. 2009).

Principle 7 Organisation models and projectsplans are fundamental artefacts.Organisation models and project plans mustbe observed as variables whose values are cap-tured during the as-is state assessment. Thisalso means that architectural views, viewpoints,models and other architectural artefacts shouldbe regarded as organisation variables. For ex-ample, the repository of a UML modelling toolholding the specification of a system underdevelopment must be an organisation artefactbecause it contributes to the specification ofthe to-be state. In contrast, a project is oftenregarded as an organisation artefact. For in-stance, both TOGAF and ArchiMate explicitlyconsider the concept of project Work Pack-age. However, organisational models, view-points and views are not explicitly regardedas artefacts by enterprise architecture model-ling languages. Nonetheless, system architec-ture guidelines such as ISO 42010 point out theimportance of considering these elements assystem artefacts (ISO/IEC/IEEE 2011).

Principle 8 The to-be state is sufficient to plana transformation project.For the purpose of planning a transformationproject the current as-is state is not requiredbecause the to-be state must fully specify theorganisational goals.

4.3 DiscussionFigure 4 depicts a time line and a series of eventsin time (T0-T5). T0 represents the current mo-ment, therefore indicating the instant the as-is



Figure 4: Project planning and execution.

state was captured. At T0 the project P is con-ceived and enters the gestating state: this projectis planned to start at T3 and to be completed atT5. Events T1, T2, T4 signal the completion ofprojects X, Y and Z, respectively. Therefore, T1,T2, T4 also indicate that the artefacts that wereproduced by these three projects became alive.Since project P is planned to start at T3 the or-ganisation requires knowing about its state atstate to-be(T3) and not at state as-is(T0) althoughplanning is actually taking place at T0. This hap-pens because the completion of projects X andY at T1 and T2 may interfere with the executionof P at T3. Furthermore, the organisation alsorequires knowledge about its state at T4 becausethe changes resulting from project Z may alsointerfere with project P.

To plan a transformation initiative an organisa-tion needs to be aware of the set of to-be stateswhile the project is being executed. A descrip-tion of the as-is state for planning purposes isactually of limited use because there is often atemporal gap between project planning and pro-ject execution. On the other hand, other projectsconclude and change the organisation state whilethe project stands between planning and execut-ing. These observations minimise the relevanceof the as-is state as a means to design the trans-formation processes of the organisation.

As an example, consider an organisation thatplans the replacement of a system in 6 months

time and starts today the corresponding projectplan. The project planning phase must have anunderstanding of the dependencies between thatsystem and other systems, as well as to the busi-ness processes it supports. If no state changesoccur in the next 6 months, then the organisa-tion can indeed rely on the as-is state to planthe replacement project. But if the organisationis performing a set of additional transformationprojects that will change the organisation’s stateduring that period, then planning the system re-placement project will require knowing aboutthe sequence of to-be states during the next 6months and during the actual execution of thereplacement project. Otherwise, it will not bepossible to plan according to the actual networkof dependencies between the system to replacedand other organisational artefacts. Therefore, forthe purpose of project planning and execution,the current as-is state will often not mirror theorganisation’s reality. In fact, the relevance ofthe as-is state is inversely proportional to thenumber of projects being completed per unit oftime. At the limit, all dependencies of the systemto be replaced may change between the planningand execution phases, meaning that all as-is statevariables will become irrelevant for planning pur-poses.

Nevertheless, the knowledge about an organisa-tion’s current state is a fundamental asset for itsoperational management. At operational level,


actions and reactions are based on near real-timeobservations and events, meaning that planningand execution occur in close sequence. However,the requirements of the near real-time opera-tional level of an organisation should not be inter-twined with the medium to long-range require-ments required for organisational transformationand governance.

5 Conclusions

Organisations do plan and execute projects, re-gardless of not having a full or accurate rep-resentation of the as-is or to-be states. Suchan accomplishment implies that projects includeto some degree an assessment of the impact ofchange between and during planning and execu-tion.

An organisation that does not have a represent-ation of its to-be state will be unable to create adetailed plan of project P as depicted in Fig. 4.This means that parts of the plan must be post-poned until T3 to minimise the gap between theplanning and execution of P. This reality is com-monly observed in many organisations despitehaving impact on the project costs and risk, andstaff assignment. It also interferes with the plan-ning of other projects, thereby having negativeimpact on the organisation’s agility. To remedythis issue, enterprise architecture projects oftenattempt to obtain a complete and accurate rep-resentation of the as-is state. As a result, theprimary goal of these projects is an attempt tokeep an organisational repository updated withan observation of the as-is state. This approach isoften justified by statements such as “knowing indetail where we stand today is a pre-requirementto any transformation project.” Although thissounds wise, this is a demanding task in terms ofeffort and time. Moreover, and as discussed be-fore, the rate of organisational change will makethe as-is state obsolete for the purpose of trans-formation planning. Therefore, we posit thatorganisations should reassess the actual value of

enterprise architecture projects that aim captur-ing the as-is state as an enabler of transformationplanning.

This dilemma is found in many organisations:the contrast between the notion that an as-is as-sessment is a valuable asset for organisationaltransformation, and knowing at the same timethat achieving such continuous task is demand-ing. This paper defends that an organisation doesnot need to have a full and accurate depiction ofthe as-is state but of its to-be state. The to-bestate is specified according to the specific goalsof projects, that are required for planning pur-poses. This contrasts with the as-is state thatrequires observing the variables of all organisa-tional artefacts that are not retired. Consider aproject that aims creating a new system that willinteract with an existing legacy system. Planningthis project requires collecting information aboutthe legacy system as well about the design of thenew system. However, the task of collecting in-formation about a legacy system for the purposeof project planning is actually contributing toextending the knowledge about the current stateof the organisation. This is a potential avenueto sort out the dilemma stated earlier because arepresentation of the as-is state can be built in-crementally by specifying the to-be state(s) thatare required to plan the multiple projects of anorganisation.

This position paper has presented a general frame-work that provides representations of dynamicorganisations in the context of enterprise engin-eering. It specifically describes a set of prin-ciples grounded on dynamic systems theory thatprovide guidelines on how to represent a carto-graphic representation of an organisation. Suchrepresentations facilitate the planning of organ-isational transformation.

References

van der Aalst W. (2011) Process mining: discov-ery, conformance and enhancement of busi-ness processes. Springer



van der Aalst W., Adriansyah A., de Medeiros A.K. A., Arcieri F., Baier T., Blickle T., Bose J. C.,van den Brand P., Brandtjen R., Buijs J. et al.(2012) Process mining manifesto. In: Businessprocess management workshops. Springer,pp. 169–194

Abraham R., Tribolet J., Winter R. (2013) Trans-formation of multi-level systems–theoreticalgrounding and consequences for enterprisearchitecture management. In: Advances inEnterprise Engineering VII. Springer, pp. 73–87

Adams S. (2009) ITIL V3 foundation handbookVol. 1. itSMF International, Office of Govern-ment Commerce, ISBN: 978-0113311972

Agrawal R., Gunopulos D., Leymann F. (1998)Mining process models from workflow logs.In: Lecture Notes in Computer ScienceVolume 1377. Springer, pp. 467–483

Andrei N. (2005) Modern Control Theory: Ahistorical perspective. Center for AdvancedModeling and Optimization, Romania

Antunes G., Caetano A., Bakhshandeh M.,Mayer R., Borbinha J. (2013) Using Ontolo-gies to Integrate Multiple Enterprise Archi-tecture Domains. In: Abramowicz W. (ed.)Business Information Systems. Lecture Notesin Business Information Processing Vol. 160.Springer, pp. 61–72

Antunes G., Bakhshandeh M., Mayer R., Borb-inha J., Caetano A. (2014) Using Ontologiesfor Enterprise Architecture Integration andAnalysis. In: Complex Systems Informaticsand Modeling Quarterly 1

Caetano A., Assis A., Tribolet J. (2012a) UsingBusiness Transactions to Analyse the Con-sistency of Business Process Models. In: 45thHawaii International Conference on SystemScience, pp. 4277–4285

Caetano A., Tribolet J. (2006) Modeling organ-izational actors and business processes. In:Proceedings of the 2006 ACM Symposium onApplied computing. ACM, pp. 1565–1566

Caetano A., Silva A. R., Tribolet J. (2009) A role-based enterprise architecture framework. In:Proceedings of the 2009 ACM Symposium on

Applied Computing, pp. 253–258Caetano A., Pereira C., Sousa P. (2012b) Gener-

ation of Business Process Model Views. In:Procedia Technology 5, pp. 378–387

Dietz J. L., Hoogervorst J. A., Albani A., AveiroD., Babkin E., Barjis J., Caetano A., HuysmansP., Iijima J., Kervel S. J. V. et al. (2013) Thediscipline of enterprise engineering. In: In-ternational Journal of Organisational Designand Engineering 3(1), pp. 86–114

Dumas M., Rosa M. L., Mendling J., ReijersH. A. (2013) Fundamentals of Business Pro-cess Management, 2013 Edition. ISBN 978-3642331428. Springer, p. 350

Engel R., van der Aalst W. M., Zapletal M.,Pichler C., Werthner H. (2012) Mining inter-organizational business process models fromEDI messages: A case study from the auto-motive sector. In: Advanced Information Sys-tems Engineering. Springer, pp. 222–237

Eriksson D. M. (1997) A Principal Exposition ofJean-Louis Le Moigne Systemic Theory. In:Cybernetics & Human Knowing 4(2), p. 42

Filipe J. A. C. B., Coelho M. F. P., Ferreira M. A.M., Figueredo I. C. (2011) Social Responsib-ility and Environmental Sustainability: TheCase of Caixa Geral de Depósitos (Portugal)and Vale (Brazil). In: Editorial Board Mem-bers 10(5), pp. 352–375

Hoogervorst J. A. (2009) Enterprise governanceand enterprise engineering. Springer

ISO/IEC/IEEE (2011) 42010. Systems and Soft-ware Engineering: Architecture Description

Lankhorst M. (2013) Enterprise architecture atwork: Modelling, communication and ana-lysis. Springer

Levine W. S. (1996) The Control Handbook Lev-ine W. S. (ed.). CRC Press

le Moigne J.-L. (1977) La theorie du systemegeneral. Theorie de la modelisation. Collec-tion les classiques du reseau intelligence dela complexite

Molka T., Redlich D., Drobek M., Caetano A.,Zeng X.-J., Gilani W. (2014) ConformanceChecking for BPMN-Based Process Models.In: Proceeedings of the 29th ACM Symposium


On Applied ComputingNegash S (2004) Business Intelligence. In: Com-

munications of the Association of Informa-tion Systems 13, 177â195

Op’t Land M. (2009) Enterprise architecture: cre-ating value by informed governance Springer(ed.). Springer

Potgieter A., Bishop J. (2003) The ComplexAdaptive Enterprise: Self-Awareness andSustainable Competitive Advantage usingBayesian Agencies, In: The Journal of Con-vergence, UCT Graduate School of Business

Santos C., Sousa P., Ferreira C., Tribolet J. (2008)Conceptual model for continuous organiza-tional auditing with real time analysis andmodern control theory. In: Journal of Emer-ging Technologies in Accounting 5(1), pp. 37–63

Sousa P., Pereira C., Vendeirinho R., CaetanoA., Tribolet J. (2007) Applying the ZachmanFramework Dimensions to Support BusinessProcess Modeling. In: Digital Enterprise Tech-nology. Springer, pp. 359–366

Sousa P., Lima J., Sampaio A., Pereira C. (2009)An approach for creating and managing en-terprise blueprints: A case for it blueprints.In: Advances in Enterprise Engineering III.Springer, pp. 70–84

Sousa P., Gabriel R., Tadao G., Carvalho R.,Sousa P. M., Sampaio A. (2011) EnterpriseTransformation: The Serasa Experian Case.In: Practice-Driven Research on EnterpriseTransformation. Springer, pp. 134–145

The Open Group (2009) TOGAF, the open grouparchitecture framework, version 9. Van HarenPublishing

Zacarias M., Caetano A., Pinto H., Tribolet J.(2005) Modeling Contexts for Business Pro-cess Oriented Knowledge Support. In: AlthoffK.-D., Dengel A., Bergmann R., Nick M., Roth-Berghofer T. (eds.) Professional KnowledgeManagement. Lecture Notes in Computer Sci-ence Vol. 3782. Springer Berlin Heidelberg,pp. 431–442

José Tribolet

Department of Computer Science andEngineeringInstituto Superior TécnicoUniversity of LisbonAv. Rovisco Pais1049-001 [email protected]

Pedro Sousa

Department of Computer Science andEngineeringInstituto Superior TécnicoUniversity of LisbonAv. Rovisco Pais1049-001 LisboaPortugalandLink ConsultingAvenida Duque de Avila 231000 [email protected]

Artur Caetano

Department of Computer Science andEngineeringInstituto Superior TécnicoUniversity of LisbonAv. Rovisco Pais1049-001 LisboaPortugalandINESC-IDInformation Systems GroupRua Alves Redol 91000-029 [email protected]


50 Jorge L. Sanz

Jorge L. Sanz

Enabling Front-Office Transformation and CustomerExperience through Business Process Engineering

The scope of business processes has been traditionally circumscribed to the industrialisation of enterpriseoperations. Indeed, Business Process Management (BPM) has focused on relatively mature operations, withthe goal of improving performance through automation.However, in today’s world of customer-centricity and individualised services, the richest source of economicvalue-creation comes from enterprise-customer contacts beyond transactions. The need to make senseof a mass of such touch-points makes process a prevalent and emerging concept in the Front- Office ofenterprises, including organisational competences such as marketing operations, customer-relationshipmanagement, campaign creation and monitoring, brand management, sales and advisory services, multi-channel management, service innovation and management life-cycle, among others.While BPM will continue to make important contributions to the factory of enterprises, the engineeringof customer-centric business processes defines a new field of multi-disciplinary work focused on servingcustomers and improving their experiences. This new domain has been dubbed Business Process Engineering(BPE) in the concert of IEEE Business Informatics.This paper addresses the main characteristics of BPE in comparison with traditional BPM, highlights theimportance of process in customer experience as a key goal in Front-Office transformation and suggests anumber of new research directions. In particular, the domains of process and information remain todaydisconnected. Business Informatics is about the study of the information process in organisations and thus,reuniting business process and information in enterprises is a central task in a Business Informatics approachto engineering processes. Among other activities, BPE is chartered to close this gap and to create a suitablebusiness architecture for Front-Office where organisational and customer behaviour should guide and benefitfrom emerging data analytics techniques.

1 Process is out of the IndustrialisationBox

Business process has been at the center of thestage in both research and industry for severaldecades. Under the brand of Business ProcessManagement (BPM), business process has attrac-ted a great deal of attention from many practi-tioners and scholars. BPM has been defined asthe analysis, design, implementation, optimisa-tion and monitoring of business processes (Du-mas et al. 2013; Franz and Kirchmer 2012; Rosen-berg et al. 2011; Schönthaler et al. 2012; Sidorovaand Isik 2010). Aalst et al. (2003) defined some tar-gets of BPM: ". . . supports business processes using

methods, techniques, and software to design, enact,control and analyse operational processes involvinghumans, organisations, applications, documentsand other sources of information."1

While the above definitions are quite compre-hensive and broad, in reality most BPM researchand industry activity has grown upon the motiv-ation of reducing operating costs through auto-mation, optimisation and outsourcing. There area several Schools of thought and practice (suchas lean, lean sixsigma, and others (Andjelkovic-

1Aalst et al. (2003) exclude strategy processes from BPM,a remarkable pointthat will be revisited in more depth laterin this paper.

Enterprise Modelling and Information Systems ArchitecturesVol. 9, No. 1, June 2014Enabling Front-Office Transformation and Customer Experience through Business Process Engineering 51

Pesic 2007; Andjelković Pešić 2004, 2006; Näslund2008)) and a myriad of related literature in the last40 years that serve to illustrate the focus on costcontention. Around the middle of the past dec-ade, T. H. Davenport (2005) stated in a celebratedHarvard Business Review paper that processeswere being "analyzed, standardized, and qualitychecked", and that this phenomenon was happen-ing for all sort of activities, stated in Davenport’sown terms: "from making a mouse trap to hiringa CEO". The actual situation is that industry in-vestment and consequential research have stayedmuch more on "trapping the mouse" than in dif-ferentiating customer services through innovat-ive and more intelligent processes, let alone hir-ing CEOs. This may be explained partly fromDavenport’s own statements in 2005: "Processstandards could revolutionize how businesses work.They could dramatically increase the level andbreadth of outsourcing and reduce the numberof processes that organizations decide to per-form for themselves" (bold face is added herefor emphasis).

With the advent of different technologies suchas mobile, cloud, social media, and other digitalcapabilities that have empowered consumers, theclassical approach and scope of business processhave begun to change quickly. Organisations areadopting new operating models (Hastings andSaperstein 2007) that will drastically affect theway processes are conceived and deployed. Asstated by many authors in the last four decades,business process work is supposed to cover allcompetences in an organisation, irrespective ofthe specific skills from human beings participat-ing in such operations. However, in an unpub-lished inspection of about 1,300 papers conduc-ted by the author and some of his collaborators2,most process examples shown in the literaturedeal with rather simple forms of coordination ofwork, mostly exhibiting a flow structure and ad-dressing administrative tasks (like those capturedin early works on office information systems).

2The co-authors are L. Flores and V. Becker both fromIBM Corporation.

Furthermore, the examples provided usually dealwith rather idealised operations, probably offeredas simple examples with the purpose of illustrat-ing theoretical or foundational research results(Aalst 2004; Aalst and Hee 2002; Aalst et al. 2003;Yan et al. 2012). Thus, radically simplified ver-sions of "managing an order", "approving a form","processing a claim", "paying a provider", "deliv-ering an order" etc. are among the most popularexamples of processes found in the literature.

The lack of public documentation of substantialcollections of real-world processes is remarkable.Houy et al. (2010) both confirmed the dominantfocus on simple business processes and also sug-gested potential practical consequences of relatedresearch: ". . . there is a growing and very activeresearch community looking at process modellingand analysis, reference models, workflow flexib-ility, process mining and process-centric service-oriented architecture (SOA). However, it is clearthat existing approaches have problems dealingwith the enormous challenges real-life BPM pro-jects are facing [. . . ] Conventional BPM researchseems to focus on situations with just a few isol-ated processes . . . ". Of course, the list of availablereal-world processes would be a lot richer if oneincluded the set defined by enterprise packagedapplications (Rosenberg et al. 2011). However,this comprehensive collection is proprietary be-cause it constitutes a key piece of intellectualcapital coming from software vendors or integ-rators in the industry.

The traditional focus on process has also raisedmuch controversy. At the S-BPM ONE Confer-ence in 2010, a keynote speaker (Olbrich 2011)remarked: "Let me be as undiplomatic as I pos-sibly can be without being offensive [. . . ] The aca-demic community is as much to blame [. . . ] as thevendors of BPM systems, who continue to reducethe task of managing business processes to apurely technological and automation-oriented level". While other authors in the sameconference debated "who is to blame" very an-imatedly (Fleischmann 2011; Singer and Zinser


52 Jorge L. Sanz

2011) it is important to highlight that the state-ment from Olbrich (in bold face above for em-phasis) reinforces that BPM has mostly followedthe obsession of automation and optimisation bymeans of Information Technology.

A detailed inspection of the extant literature con-firms that business process work has been de-voted to a rather small fraction of the actualvariety and complexity found in enterprise be-haviour. This behaviour enacts many valuegen-erating capabilities that organisations cultivatebased on skills provided by their own workforcesand through rich interactions with other enter-prise stakeholders, particularly customers. Thefollowing points offer a simplified summary:

(1) Business process research in Computer Sci-ence has been traditionally focused on certainclasses of enterprise operations, mostly involvingsimple coordination mechanisms across tasks.This type of coordination and the overall beha-viour represented in underlying models reflectvery much an "assembly line" where work is lin-early synchronised to deliver a desired artifactor outcome. BPMN, emerged from OMG as theindustry standard for business process modellingis a good illustration of this point. Simplicity ofthe choreography is ensured by removing anyform of overhead in communication when mov-ing from one stage to the next. Unlike other morecomplex business processes, many software ap-plications do have this simplified structure. Infact, a trend since the early 2000’s is to separ-ate the specific application logic from the co-ordination / choreography needed across mod-ules, and both of them from the actual data con-tained in a data-base management system. Dif-ferent foundations and a plethora of languageshave been created to capture this semantics ofcoordination such as Business Process ModellingNotation (BPMN), Business Process ExecutionLanguage (BPEL), Unified Modelling Language(UML), Event Process Chain (EPC), Petri Nets,etc.

(2) Resulting process models have typically yiel-ded the form of a "workflow" (Sharp and McDer-mott 2009; White 2004). This means that the activ-ation of a task in the assembly line only occurswhen certain predefined events take place, one ormore previous tasks are completed and their pro-duced artifacts transferred to the next task in thepipeline for continuing "the assembly". In fullyautomated systems, like software applications,this is a good abstraction (see Fig. 1). On theother hand, in actual business processes wherehumans participate or supervise the individualtasks, workflows do not always capture the ac-tual pattern of work, including the contractualcommitments made across role-players.

Consequently, IT systems used to implementsuch workflows, called "Business Process Man-agement Systems" (BPMS) in IT jargon3, are notsuitable to communicate the nature of work tobusiness stakeholders. This point has been ex-tensively addressed in recent Enterprise Engin-eering work (Dietz et al. 2013), such as DEMOand related contributions (Albani and Dietz 2011;Aveiro et al. 2011; Barjis et al. 2009; Proper etal. 2013). The issue of clarity was brought upby Dietz eloquently during a key-note entitled"Processes are more than Workflows" in the 2011KEOD Conference: "With modelling techniqueslike Flowchart, BPMN, Petri Net, ARIS/EPC, UMLand IDEF you get easily hundreds of pages of pro-cess diagrams. Nobody is able to understand suchmodels fully. Consequently, nobody is able to re-design and re-engineer a process on that basis".

Beyond communication issues, the distinctionof contexts between an organisational design

3The term BPMS is somewhat questionable because itimplies that these IT systems implement processes whilethey actually do so only for very special types of processes,i.e., workflows. Thus, the earliest denomination of Work-flow Management Systems (WMS) is more adequate. Asan example, Cases emerged later in the software industryand model more complex processes. The term Case Man-agement Systems (CMS) has been used to distinguish themfrom BPMS. This incorrectly implies "cases are not businessprocesses".


concern and an IT concern should also be care-fully addressed. In the workflow abstraction, thepotential role-players assigned to the executionor supervision of the individual tasks will be"idling" unless they get activated through thepipeline. This model of reality is well-suited tofully automated tasks (like those realised by soft-ware) but unsuited to other situations in organ-isations where humans take part of the processexecution.

Appl

icat

ion

1960

Appl

icat

ion

databasesystem

Appl

icat

ion

user interface

database system

Appl

icat

ion

BPM sy

stem

user interface

database system

1975 1985 2000

Appl

icat

ion

1960

Appl

icat

ion

databasesystem

Appl

icat

ion

user interface

database system

Appl

icat

ion

BPM sy

stem

user interface

database system

1975 1985 2000

Figure 1: The evolution of information systems devel-opment and the role of BPM systems in the newestgenerations of software (from Aalst et al. 2003).

Indeed, the factory model of operations capturedinto a workflow implies that people are actually"doing nothing" unless their "activation" occursby the preceding tasks in the pipeline. The latteris far from modelling accurately the reality ofwork in most enterprise processes.

(3) The tradition of business process managementworks on the assumption that the investmentmade in optimally designing a process will berecovered through the repeated application ofthe process for a long-enough period of time.The principle is that economic benefits will ac-crue from accumulated cost reduction obtainedby the application of the optimised process overand over again. This approach reflects a true’factory’ in the conception and modelling of or-ganisational behaviour. Furthermore, the idea ofperfecting the process with such an effort pay-ing off through hundreds of thousand repetitions

or even millions of interventions done with thesame process is adversarial to the business needof introducing modifications. As organisationshave been progressively more affected by suddenchange or involved in operations where changeis a common requirement this type of factoryoptimisation does not work. In fact, rigidity ofprocess models has been a long-standing and bit-ter finding. More recently, the broader issue ofprocess evolvability in the presence of continu-ous change has been the subject of solid research,including a recent PhD thesis (Nuffel 2011) andreferences therein.

(4) Implicitly or explicitly in the traditional ap-proaches to business process, it lies the Taylorianprinciple of replacing individuals by applyingautomation whenever possible. As in other busi-ness theories that build on a "dehumanisation" ofenterprises, the consequence is that the role ofhumans as sources of value-creation in processesis ignored. The connection of this foundationand BPM work has been openly recognised byVan der Aalst in his recent review of a decade ofBusiness Process Management conferences Aalst(2012): "Adam Smith showed the advantages of thedivision of labor. Frederick Taylor introduced theinitial principles of scientific management. HenryFord introduced the production line for the massproduction of black T-Fords. It is easy to see thatthese ideas are used in today’s BPM systems".

In close connection to this moral coming fromcertain economics and business schools, it alsoresides the goal of avoiding variation of the pro-cess by all possible means. This good idea origin-ally coming from manufacturing practices (i.e.,reducing variation as a means to controlling qual-ity and cost of the resulting production) has beentranslated to other forms of operations (such asservices) where variation is inevitable when inter-action with non-automated agents becomes anintegral part of the actual production process.4

4Most call centers begin all their interaction with cus-tomers by following pre-established routines. In some cases,this may disgrace the effectiveness of the service and satis-faction of the caller. A known example is when reasonably


54 Jorge L. Sanz

Inevitable process variation is a significant signof ’lost control’, as organisational capabilities gofrom the tangible to the less tangible. As said inLe Clair (2012), the less tangible the capability, themore control will be ceded to the customer. Thetradition of BPM work contrasts sharply with En-terprise Engineering (Dietz et al. 2013), a theoryin which humans are seen as a precious source ofvalue, particularly for achieving improvementsand differentiation. In particular, all processesinvolving interaction with customers offer thisopportunity (services researchers often call thisconcept "co-creation").

(5) It is important to recall that existing processclassifications such as the Process ClassificationFramework (Process Classification Framework(PCF)) reveal common areas of work in organisa-tions that do not follow the BPM tradition in thesense that they do not represent work amenableto workflows. Indeed, PCF is a standardisationeffort in different industries that includes manynon-factory areas of an enterprise. Consequently,these operations are not adequately addressed bythe application of existing BPM research, meth-ods and tools.

The clarification from Van der Aalst and his col-laborators when excluding strategy processes fromthe scope of their work was an excellent andvery early sign, although "strategy" should nothave been the only area excluded from the scopeof their contributions. Indeed, there are othercritical business processes in enterprises beyond"strategy" that do not fit workflow models, PetriNets, BPMN, or related instruments popular inComputer Science (Sanz et al. 2012). Specifically,these other forms of organisational behaviourbeyond ’the factory’ involve complex activitiescarried our by humans in collaboration with oneanother and with the support of technology inways that are observable and may also be cap-tured into process models. This point can also

educated customers are asked first whether their obviouslynonfunctioning product is plugged to the power supply, tounplug and plug it again, try to turn it on once more, andso on.

be easily illustrated by using some of the ProcessClassification Framework (PCF) content.

While some people may argue that this frame-work may arguably be called a process architec-ture (Eid-Sabbagh et al. 2012; Miers 2009; M. A.Ould 1997) it still provides a solid clue of manyoperations that are either common across indus-tries or unique to specific industry segmentssuch as retail banking or consumer packagedgoods. None of these enterprise operations canbe modeled by workflows.

In addition, the componentised business architec-ture and its resulting industry models addressedin Sanz et al. (2012) are also very useful to illus-trate the same points. In these approaches, thereis no functional decomposition at the heart of themodelling, unlike in PCF, and thus the resultingconstruction follows more closely some of thecore principles of Enterprise Engineering (Dietzet al. 2013). This will be addressed briefly in thenext section.

(6) Another important evidence that process hasmoved out of the industrialisation box is CaseManagement (more recently also called AdaptiveCase Management by the authors in (Swensonet al. 2010) and Dynamic Case Management byanalysts in Forrester). The need for Case Man-agement has been illustrated with different en-terprise operations such as claim processing inProperty and Casualty Insurance, customer ap-plications in Social Services, Health Care claimprocessing, Judicial Cases, and so on. Van derAalst and others (Aalst and Berens 2001; Aalstet al. 2005) presented Case Handling as a newparadigm for supporting flexible and knowledgeintensive business processes. In his work on casemanagement, De Man (2009) states that ’work-flow’ is an adequate representation for factory-type, highly predictable behaviour admitting forlittle or no deviation from pre-established models.In recent literature (Khoyi 2010), the argument insupport of the need for Case Management hingedaround the fact that "Case Management allowsthe business to be described in known terms ratherthan artificially fitting it into a process diagram".


2 © 2009 IBM Corporation

CUSTOMERS & PROSPECTS

Online

Email

Mobile

Call center

Branches

Kiosk, ATM, etc.

Point of sale

Postal mail

strategy

data

tech

strategy

data

tech

strategy

data

tech

strategy

data

tech

strategy

data

tech

strategy

data

tech

process

strategy

data

tech

strategy

data

tech

SILOS

SILOS

SILOS

SILOS

SILOS

SILOS

SILOS

SILOS process

process

process

process

process

process

process

1 © 2009 IBM Corporation

CUSTOMERS & PROSPECTS

Online

Email

Mobile

Call center

Branches

Kiosk, ATM, etc.

Point of sale

Postal mail

Enterprise

One seamless entity

behind all the channels

Figure 2: Customers and prospects deal with an enterprise through a number of channels by following patternsor Journeys that vary according to individuals’ goals and behaviour. The picture on the upper side represents theexpected experience of meeting the enterprise as a single and well-integrated entity. However, reality is very differentas channels are not well-integrated, represented visually by the horizontal silos of the lower side picture. Each silohas their own processes, data, strategy (incentives) and IT. Thus, a Customer Journey is the integration of individualcustomer-enterprise touch-points to realise a specific customer outcome. These Journeys are essential processes deeplyrelated to loyalty and other significant measures of customer experience, unlike traditional customer satisfaction metrics.


56 Jorge L. Sanz

2 Process and the Broken CustomerExperience

In the context of this paper, customer experienceis the conjunction of all experiences a consumerhas with an enterprise over the duration of theirrelationship (Harrison-Broninski 2005). Customerexperience is critical for enterprises because ithas been widely understood as a key factor driv-ing customer loyalty (Propp 1968). Poor customerexperience in business-toconsumer enterpriseshas been a top concern in organisations for longerthan five years. The main reason is the pro-found lack of loyalty that customers exhibit inthe business-toconsumer (b-to-c) industries (Cap-gemini 2012).5 While this challenge has been com-monplace in many industry segments, the prob-lem is particularly acute in most b-to-c servicesorganisations where many initiatives have beentaken to address the problem, even to the pointof introducing a new role at the top managementlevel called Customer Experience Officer (Bliss2006).

The advent of multiple channels of engagementfor the same enterprise exposes deeper gaps inthe way organisations deal with their custom-ers. Indeed, multiple channels have generatedeven more disconnects with customers as thesechannels are generally managed by different or-ganisational units and have isolated measuresof performance. Traditional customer satisfac-tion measures tend to focus on individual cus-tomer interactions on a specific channel but thesedo not seem to correlate positively with cus-tomer loyalty (Rawson et al. 2013; Stone andDevine 2013). Figure 2 illustrates customer in-

5In North America, 80% of clients are "happy" with theirbank service but only 50% say they will remain with theircurrent bank over the next 6 months. This reflects thefinding that globally, only 42% of bank customers haverate their experience as being positive. Furthermore, sat-isfaction levels with branches, despite being the most ex-pensive and most developed channel, averages 40% world-wide with highest being 60% in North America (Capgemini2012).

teractions6 taking place across different enter-prise channels (upper side of the Fig.). Thesepatterns are typical for a single customer pursu-ing a specific outcome. In most organisations,each channel behaves as a silo (lower side ofthe Fig.) thus having its own strategy, goals,processes, data and technology. This discon-nect across channels impacts customer experi-ence quite negatively. In summary, a much moreengaged consumer through multiple channelsis making the already disrupted customer ex-perience unmanageable for large enterprises.

All these challenges lead organisations to revisitsome of their core competences related to cus-tomer experience. In fact, a number of key capab-ilities have been emerging over the last decade,starting to yield best-practices for front-office op-erations (Hastings and Saperstein 2007). However,it is the lack of understanding, modelling andinstrumenting critical customer journeys themain reason why customer experience contin-ues to be disrupted and has got worse with theadvent of more channels. Furthermore, aligningthese customer journeys with back-office opera-tions yielding end-to-end business processes is es-sential to enable customer experience. Businessanalysts characterised this new process trend dir-ectly affecting customer experience under dif-ferent names and also alerted practitioners, re-searchers and process professionals about dif-ferent shifts taking place along the entire "hypecycle" of process evolution. In particular, Forres-ter used the name "tamed processes" and char-acterised them as follows: "Tamed processes aredesigned from the outside in, can be driven bybig data and advanced analytics, support socialand mobile technology, provide end-to-end support

6The set of customer-enterprise interactions followed toachieve a specific outcome for an individual costumer hasbeen named customer journey (Rawson et al. 2013). This termhas probably been coined by some technical and businesspeople with the goal of implying that the concept should notbe made part of the classical "process grinding" experiencedthough four decades of BPM, lean six-sigma and the like.Beyond communication intent, journeys are processes andthis is a well-supported fact in Social Science work.


across systems of record and functional areas, andlink on-premises and cloudbased services" (Le Clair2012).

Engineering (i.e., designing and running) thesecustomer journeys is a very different problemfrom those BPM has been focusing in four dec-ades. These needs around modelling and archi-tecting for customer experience are in sharp con-trast to applying Customer Relationship Man-agement (CRM) packaged applications used tomonitor sales, manage customer center calls ordesign optimised workflows for efficient backof-fice processes. In fact, there is a risk that soft-ware may be used precipitately for supportingenterprise capabilities related to customer exper-ience. Indeed, some of these emerging practicesare being made into software without adequateexposure of the underlying business processes.This should constitute a warning to managementas these software applications bury rich busi-ness processes into their packaged software,thus signaling the same issues experienced inmature back-office operations. This warning isa significant call for the adequate research andpractice necessary to surface the key processesbefore they are fully embedded into "concrete",a fact that will impact agility as the frequencyof change in these processes is a lot higher thanin those modeled in conventional enterprise re-source planning. Traditional approaches to busi-ness process instrumentation based on packagedapplications in conjunction with custom BPMsystems come to memory after four decades ofcost-take out and efficiency improvements. Inpart, this rigidity has created fragmented cus-tomer experience as a consequence of the lackof flexibility and long time-to-value for desiredchanges in the information technology systemsdeployed across the enterprise. This is an obser-vation coming from direct practice in the fieldand can also be corroborated by exploring a veryextensive business literature. In short, if front-office processes are not addressed according tothe new business and societal needs, the on-going fragmented experience will result in ad-

ditional loss of loyalty and consequently, cus-tomer equity or profitability issues (Villanuevaand Hanssens 2007).

Probably to the surprise of many data analyt-ics advocates, if customer-centric processes arenot engineered to reflect the demands from thenew economy, the emphasis on individualisingcustomers and "inferring their behaviour" willjust make customer experience even worse. Thereason is that customers will increase their ex-pectations for personalised services while theability for organisations to address this expecta-tion remains far from the current state-of-the-art.This issue will become particularly challengingfor some services industries because (i) such per-sonalisation may not be viable due to the natureof the service being delivered; (ii) personalisationrequires in many cases a co-created design anddelivery, a pursuit that many enterprises are notyet in a position to address; (iii) regulatory lim-itations may prevail thus limiting the enterpriseto discriminate across customers; or (iv) scalab-ility of good quality customer service may be atodds with profitability targets. This remark is anattempt to warn "data scientist" approaches tofront-office operations, as the main disconnectswill only be widened by "data-only" insights.

3 Process in critical areas of theFront-Office

The term "Front-Office" is used here to denotethe set of enterprise activities and resources ded-icated to the support of customer experience. Inthis category, they fall many customer servicemanagement operations. But other Front-Officeareas in organisations also go beyond the pur-pose of dealing directly with customers. Someexamples are brand monitoring, campaign designand deployment, enterprise marketing operations,product and service innovation, customer loyaltyand advocacy management and others amongthe top areas where organisations have been in-vesting in the last decade or so. These enterprisecapabilities and related competences support cus-tomers indirectly, although boundaries may blur


58 Jorge L. Sanz

in some cases (for example, a campaign designmay involve realtime intervention based on cus-tomer interactions). These capabilities are be-ginning to have more visible best-practices andthus, corresponding business processes are emer-ging. Consequently, their study is at the realm ofBusiness Process Engineering because they en-compass key work-practices. These operations in-volve humans and collaborative activities deeplyinterrelated with technology and information,and their patterns of work are also emerging,become more and more visible, being subjectedto white box modelling rather than remaining asblack boxes. In these new process areas, Inform-ation Technology will still be essential but inradically different ways from "the factory" of en-terprises. Actually, translating those experiencesfrom Information Systems in the Back-Office tothe Front-Office is a sure recipe for disaster. Thisinadequate translation would also add significantlongterm strategic and cost-centric consequencesto the ongoing broken customer experience.

Searching for further practical evidence on theemergence of non-traditional enterprise areasneeding process study, it is important to revisitin depth some theories of organisational designand related work by different business researchschools (Penrose 2009). Figure 3 shows an or-ganisation of the resource-base of a typical en-terprise into four distinct types and the corres-ponding bundling of such resources into disjointbusiness components. Each column on the righthand side of the Fig. represents one typical com-petence whose organisation is described by thegeneric concepts of the column on the left, aspresented in Sanz et al. (2012). Although a dif-ferent language was used, the foundations ofthe structure of a generic competence should behonored to Brumagin in Brumagim (1994), amongother more recent business researchers.7

7This is probably the only known actionable model de-rived from the general and powerful concepts running un-der the denomination of Resource-Based View (RBV) inthe theory of the firm. Business process researchers arestrongly encouraged to delve into RBV, search for cross-pollination with related Social Sciences work, and revisit

Notice that the hierarchy of resources represen-ted in Fig. 3 does not mean the same as the clas-sical management concept of "control". Instead,it only represents an arrangement in which dif-ferent skills, information, assets (intangible andcapital) and derivative entangled capabilities arebundled together to produce one or more relev-ant outcomes in the enterprise. Likewise, thesecomponents are not necessarily aligned with tra-ditional Lines-of-Business and do not intend tomap departmental capabilities or other conven-tional "reporting structures" in enterprises. Revis-iting Penrose (2009), the components highlightedon the right may be thought as the formalisedgrouping of resources whose entanglement pro-duces those core services (internal or external)that the organisation needs to serve all stake-holders. Some enterprises may be endowed withsome of these resources in unique ways, beingalso more idiosyncratic for some industries thanothers.

Concrete models recently built for many industrysegments by following the modularisation prin-ciples reveal that there are hundreds of busi-ness components that the business process tra-dition has failed to address. In fact, most pro-cesses available from the research literature fallin the category of operations involved in thelast row of business components, i.e., productionand maintenance processes. As the level of in-volved resources moves into oversight and man-agement, several interesting examples of casesmay be found and used to illustrate the typeof operations at play. Going further into learn-ing and innovation, traditional contributions fadequickly or disappear entirely. Interestingly, thetop row of Fig. 3 includes the ’strategy processe’that Van der Aalst and collaborators explicitlyexcluded from their foundational work in theearly 2000’s. A diversity of processes like thoseneeded for controlling the quality of a cartoon inan entertainment industry enterprise, managingthe pipeline of compounds in a pharmaceutical

business research topics such as those addressed in Organ-isational Behaviour schools.


Figure 3: The four types of resources defined according to the different forms of behaviour that are observed in a genericenterprise (left). Componentised organisation of such resources based on different competences (right). Each of thesecomponents deals with a number of core subjects (Nandi and Sanz 2013) whose evolution is key for the definition ofcorresponding competences (columns in the picture)

company, and disseminating the learning harves-ted from a specific family of consulting practicesthroughout a services enterprise should not be in-cluded under the term ’strategy’. However, theseoversight and management processes have notbeen addressed by the BPM tradition.

There may be still an argument that processes inclassical BPM work aim at modelling operationsacross the components and not inside them, i.e.,end-to-end processes also called ’value streams’in some business literature. However, this argu-ment does not necessarily follow from inspectingthe work reported in more than one thousand pa-pers in the last twelve years. The BPM traditionhas adequately responded to the need of min-imising transaction costs across the enterpriseand builds upon existing governance mechan-isms defined as true systems of control alignedwith functions (Le Clair 2012). In that sense thetraditional approach has followed closely the en-terprise disconnection and rigidity leading tothe present state-of-the-art in customer exper-ience. Moving the foundational basis to addressthe next generation of business process (called"hybrid connected processes" in Le Clair (2012)),crossfunctional and complex processes (i) can-not be made or realised into workflow structures

and (ii) new languages are needed to close theremarkable communication gap left in the cross-enterprise process space. It would be impossibleto address these statements in full detail herebut it should suffice to say that loss of visibilityin cross-enterprise processes is a proven pain-point (Nandi and Sanz 2013) still yielding well-identified performance and communication prob-lems in many firms. In other words, the "hundredof pages" alluded by Dietz (2011) are real and theinsight that these many pages have unraveled isminimal.

From a research perspective and practical pointof view, the reader is referred to the recent workin Nandi and Sanz (2013) for evidence that themain ’value streams’ across an enterprise arein fact progressions of core subjects and not life-cycle of objects, at least when the latter is un-derstood in the tradition of statemachines, i.e.,artifacts evolving through a number of micro-states that separate the initiation and comple-tion of "tasks". This fact goes back to the funda-mental way metaphysics of processes has beenapproached in Social Sciences (Rescher 1996) andthe conceptual duality between process and sub-jects8 in the organisation of the world of a gen-

8The word "subject" here means "theme" or "topic". This


60 Jorge L. Sanz

eric enterprise. Indeed, subjects are higher-levelabstractions than conventional objects and theirevolution is thus subjected to lots of asynchron-ous activity taking place across the enterprise.The delivery of outcomes produced by these asyn-chronous activities signals the completion of ne-cessary results as agreed in pre-determined cross-functional commitments. These commitments are,in fact, a form of organisational contracts andmay be regarded as quite granular macro-statesin the evolution of an individual subject. These’states’ are called milestones in Nandi and Sanz(2013).

The need for aligning the research agenda in pro-cess to the main challenges faced by industrywas also called out in the closing recommenda-tions from the BPM study in Indulska et al. (2009):". . . despite being an actively researched field, anec-dotal evidence and experiences suggest that thefocus of the research community is not alwaysaligned with the needs of industry". A couple ofyears have elapsed since related papers were pub-lished but the situation has not changed much.Reijers et al. (2010) also addressed the import-ance of rooting BPM activities in industrial prac-tice and correctly questioned the understandingof the actual adoption of BPM by organisations:". . . it may come as a surprise that contemporaryinsights are missing into which categories of organ-izations are adopting BPM and which type of BPMprojects they are carrying out". Actually, Aalst(2012) did some justice in his recent review ofresearch in the last decade of BPM Conferencesand highlighted that this work mostly addressedautomation concerns. In particular, Van der Aalstrevisited BPM systems as an opportunity to fur-ther position BPM tools as valuable instrumentsto build better software applications.

While this traditional BPM research work andpractices should definitely continue, new markettrends and needs from new enterprise capabil-ities in the Front-Office strongly suggest that

differs from the interpretation of subject as an actor carry-ing out an activity, and thus, it should not be confused withrelated semantics in S-BPM.

business process focus has to shift in order tocontribute to other urgent goals in organisations.Business process is called to play as a key in-strument for achieving the customer experienceneeded in front-office operations and deep end-toend integration of the latter with the back-officein enterprises. The main motivation for the newwork needed does not hinge around cost reduc-tion, industrialising routine operations or build-ing better software with BPM systems.

4 Back to Process Foundations

The evolution of business process has nothappened without significant divergence and tosome extent, also confusion. The state-of-the-art is plagued by language chasms, cultural silosand idiosyncratic viewpoints. Some of these chal-lenges were documented in De Man (2009); Indul-ska et al. (2009); Recker et al. (2009); Reijers et al.(2010) and others. Reijers et al. (2010), state thechallenge in clear terms: "Considerable confusionexists about what Business Process Managemententails . . . ". Indeed, the definition of businessprocess is still troubled by ambiguity and addingthe term "management" has done little to clarifythe confusion. A plea for this clarity has beenarticulated by Olbrich (2011): "It seems a pity thata lot of current research fails to provide a basicdefinition of what underlying understanding of’process’ and ’BPM’ it bases its work on". In fur-ther exchanges in the same S-BPM conference,other authors such as Fleischmann (2011); Singerand Zinser (2011) agreed that the problem goesfurther into a lack of clarity on the very defin-ition of BPM. A review of the literature showsthat there is not a single and agreed definitionof these terms. While ". . . a scientific foundationis missing" was clearly stated by Van der Aalstback in 2003, the review of BPM Conferences pub-lished by the author a decade later confirms thatthe fundamental shortfalls have not been over-come yet (Aalst 2012). The underlying reasonis deeply related to the nature of business pro-cess being a socio-technical system and thus, its


complexity cannot be approached by a narrow fo-cus on technology dimensions. In Fleischmann’sown words: ". . . sociological systems like organ-izations are combined with technical systems likeinformation and communication technology. For aholistic view of business process management wehave to consider all aspects" (Fleischmann 2011).Weske (2012) also highlights the deep nature ofprocess: "a business process consists of a set ofactivities that are performed in coordination inan organizational environment. These activitiesjointly realize a business goal." While using differ-ent language, other authors also defined businessprocesses (Davenport 1993; Debevoise 2007; Du-mas et al. 2013; Indulska et al. 2009; Krogstie et al.2006; Ould 1995a; Smith and Fingar 2007 and thelist goes on).

The Object Management Group recognised thefoundational problem with the definition of pro-cess. Siegel (2008), the leader of the BPM groupstated: "there is no agreed-upon industry defini-tion of Business Process. Instead, there are multipledefinitions, each looking at the field from its ownunique point of view, concentrating on its own setof concerns". Certainly, it is not a matter of onedefinition being right and the others being wrong.Rather, the issue is about the varying points ofview used. As a consequence, the main efforts inprocess modelling standardisation have not yetyielded the expected outcomes, as discussed inRecker et al. (2009), more broadly exposed in In-dulska et al. (2009) and highlighted in Aalst (2012).Unquestionably, most people do have a similarand informal notion of "business process". Butthis intuitive agreement does not mean a conver-gence across viewpoints. In fact, the variationsin the definition of process may suggest that theterm is a boundary object across disciplines, in-dividuals from different units of an organisationor communities of practice. Other researchersin Social Sciences and Philosophy have also fo-cused extensively on the concept of process andits definition. Ven (1992) addressed the topic inthe context of one of the most complex typesof processes in organisations, i.e., the strategy

process. The depth of Van de Ven’s classificationreveals the foundations underlying many busi-ness process definitions. In spite of having beenpublished two decades ago, this work has goneunnoticed in most of the BPM literature (Aalst2012; Aguilar-Savén 2004; Klein and Petti 2006;Ko et al. 2009; Lu and Sadiq 2007; Ould 1995b;Propp 1968; Toussaint et al. 1998; Trkman 2010and many others).

Another language chasm across different schoolsof thought or communities of practice is the un-clear relationship between the concept of busi-ness process and that of organisational routine.Rich literature is available on the study of routines(Becker 2004), the significance of routines as aunit of analysis for organisations (Levin 2002;Pentland and Feldman 2008; Pentland et al. 2012)the collectivist meaning of routines and the needfor establishing solid micro-foundations (Felinand Foss 2004) among others. It is very likely thatbusiness process and routine address identicalconcerns in organisation theory; however, inspite of the prolific technical production in thetwo subjects during decades, their formal rela-tionship and the reasons for keeping two differ-ent terms remain unclear.

More recently, there has been a fundamentalpiece of work in process that builds upon a recon-ciled view of process and information availablesince the early days of the Information Engineer-ing schools in Europe. This approach to businessprocess goes under the brand of entity-centricoperations modelling (Sanz 2011) and offers a hol-istic approach that reunites different types ofprocesses under the same conceptual understand-ing. This entity-centric concept has been usedintensively by (Ould 1995a,b) and although thenotion of life-cycle is from the early 1980’s, sev-eral important contributions has been made indifferent industries and software to merit a de-tailed inspection in Business Process Engineering(Bhattacharya et al. 2009; Cohn and Hull 2009;Nandi 2010; Nigam and Caswell 2003; Robinson1979; Rosenquist 1982).


62 Jorge L. Sanz

Quite interestingly, another related approach wasrecently presented to model cross-functional end-to-end processes in enterprises based on the no-tion of subjects and nexus of commitments (Nandiand Sanz 2013). The foundation for all this workappears as an important step toward the designand construction of different process types, in-cluding the so-called value streams, by using acommon approach in which information doesnot take back seat as a mere "after-thought" inthe modelling of behaviour or becomes confusedwith "state model", being the latter a commonmisunderstanding incurred by most computerscientists as Van der Aalst remarkably noted. Thepoint of reunion of these seemingly related mod-elling techniques does not reside in "artifacts"or "object life-cycle" but instead, it goes back tothe Social Sciences in the sense that the unifyingconcept is the very epistemology of process, i.e.,"things in the making" (Tsoukas 2001; Tsoukasand Chia 2002). Consequently, process design isabout describing the evolution of a core subject.While the roots of this approach come from sev-eral decades of work and different schools ofthought, not all process researchers and practi-tioners seem familiar with these concepts andrelated literature sources.

5 Research topics in Business ProcessEngineering

It would be difficult to propose here a completeagenda of research and practice in Business Pro-cess Engineering. Like in any other emergingfield of work, only the pass of the time, com-munity activities and market consolidation willdetermine its boundaries and shape its ultimatepriorities. However, based on current work andongoing industry needs, it would be safe to high-light some important areas with the purpose ofstimulating further research.

This is a first pass through such a list. Topics areclassified according to four basic categories:

Customer-Enterprise Behaviour: Foundations andModels

(A) Establish a foundation for understandingand modelling the journeys that customers fol-low in their multiple touchpoints when inter-acting with enterprises across different chan-nels. These journeys are probably the mostlooselycoupled type of processes, i.e., they arehighly unstructured but they are not "randomwalks" at all as customers seek for specificoutcomes. This type of interactions is alsofound in other collaborative work in enter-prises (Harrison-Broninski 2005). In addition,as involved interactions combine and altern-ate human-to-human and human-todigital con-tacts, these journeys are rich in informationand behaviour. Then, their adequate under-standing is imperative for the next generationof customer experience. Some work has beendone on this topic but there are no foundationsyet with a theory that explicates the journeysand how behaviour of the actors should beguided from footprints of customer contactsand previous experiences. This is one of themost fundamental research problems that dif-ferent industries need to benefit from as itsvalue is directly related to customer loyalty.

(B) Discover customer-enterprise co-creationmechanisms and have them reach a massivescale through innovative processes. This willsupport the social transformation necessaryfor the information coming from social datato become a trustable source of actual beha-viour and intent of individuals. While so-cial media means a flood of useful data, in-ferring human intention and behaviour fromthese sources remains illusory. Co-creation pro-cesses deploying collaborative and mutuallybeneficial practices appear essential for thenext generation of customer experience. Ex-plicit provision of knowledge on an individualcould be then done in exchange for personal-ised services or some other form of tangiblevalue-propositions. This will lead individualsto provide trustworthy evidence of their beha-viour and intent. Designing and implement-ing the necessary processes to reach the scale


needed requires deep socio-technical innova-tion. These processes will also help encouragefull transparency from consumers and enforceaccountability from companies. The latter willhelp replace today’s legal disclaimers in whichconsumers are asked to resign their privacyrights under terms-and-conditions that prob-ably few consumers read and even a fewernumber of them understand.

(C) Create a "sociology of the customer" thathelps understand the effect of using mass pro-cesses even with individualised clients in thepursuit of ’profitability’. If economic analysisrenders it viable, data footprints left by con-sumers will not be the only hint to infer cus-tomer behaviour (which is an erroneous ap-proach to understand people’s needs and trueexpectations anyway). Furthermore, the integ-ration of process and big data will allow forfull operationalisation of "insight", thus mak-ing the latter move from "interesting discov-ery" to a Social Science-supported theory toenhance services and provide enterprises withhigher customer equity.

Front-Office Business Architecture

(D) Propose complete Front-Office operationalmodels that represent the actual work enter-prises do with and for their own customers.This should include process and performanceframeworks for all those key competences andcapabilities in the enterprise that belong to theFront- Office operations. In particular, the cre-ation of solid Process Reference Architecturesfor emerging operational areas in marketing,brand management, campaign management,etc. would be critical for accelerating industryvalue of new research. As suggested earlier inthis paper, surfacing and documenting thesenew workpractices is essential. Software pack-ages are already in the market and these ap-plications bury important processes whose fre-quent change is imperative for flexibility ofFront-Office operations.

(E) Reconcile the ever-deepening silos of Inform-ation and Process. As suggested by the differ-ent levels shown in Fig. 4, the information andprocess domains have traditionally evolved inalmost complete isolation from each other. Asdamaging as this disconnection may result forthe well-being of any organisation, the prob-lem has stayed unresolved throughout severaldecades. In fact, the gaps have widened andgot deeper as the new "business analytics"trend has been getting momentum in enter-prises and gathering the attention of the ChiefMarketing Officer. The introduction of "bigdata" and other marketing concepts in Inform-ation Management technology has continuedto widen the chasm. Hopefully, by building ona new foundation where the Information Pro-cess in organisations and society is repurposedas a single phenomenon through Business In-formatics, new bridges will be built across thetwo silos. This reunion is dubbed "Deep Pro-cess meets Business Analytics" on Fig. 4. Theneed for this integration will reposition "pro-cess analytics" as the integration of on-line(real-time) analytics and customer journeys.

(F) Provide data-only analytics and related stat-istical modelling with a better foundationthrough behaviour-based causation. Thisshould help foster a blended approach through"white box" Enterprise Engineering modellingfor today’s decision-making techniques basedon "black-box" statistics. Among other areas ofcritical enterprise value, this topic should alsohelp define an enterprise business perform-ance framework that integrates behaviour anddata in organisations. This goal correspondsto achieving the important integration shownin the top level of Fig. 4.

(G) Develop a theory of Process Modularisationthat is consistent and evolvable with change.This work has been initiated by different col-leagues in (Nuffel 2011). As the "unit of change"in Process gets progressively more clear, thetopic of Process Evolvability will also becomeconnected to modularisation, thus addressing


64 Jorge L. Sanz

Scorecards

Big Data Massive Analy5cs

Analy5cs Capabili5es for LOBs

(Transac5onal)

Business Intelligence (from ERPs)

Data

INFORMATION SILO

Value-‐Driven Process

Management

Opera5onal Process Innova5on

Con5nued Process Improvements and

Flexibility

Cross-‐Func5onal Processes

Process Flows and ERPs

PROCESS SILO

Enterprise Performance Management

Intelligent Business Opera5ons and

Systems

Adap5ve / Agile Opera5ons

Business Visibility Modeling (BVM)

Business En55es Modeling &

Execu5on (BELA)

Process and Business Analy<cs

Integrated

Deep Integra+on

Figure 4: Silos in information and process management have deepened with the evolution of each domain. This gap ismore notorious after the advent of business analytics, scorecards, performance management and value-driven processBPM

the need for managing combinatorial effects(as already addressed by the general principlesof Normalised Systems Theory in (Mannaertand Verelst 2009) for the case of software sys-tems).

(H) Clarify the distinction, if any, between theSocial Science concept of organisational routine(Pentland et al. 2012) and the broader meaningof process coming from Business Process En-gineering. This will help reconcile work acrossthe different schools of research in Social andComputer Sciences. While practitioners sel-dom use the word "routine" (and when theydo, they imply repetitive or boring tasks whichis not the meaning in Social Sciences), it isimportant to benefit from cross-inseminationbetween Enterprise Engineering and Social Sci-ence research for better understanding of or-ganisational design through deep behaviourresearch.

(I) Benefit from Enterprise Engineering princip-les to reposition the role of humans in thevalue-creation of Front-Office business areas.This topic has several deep social connota-tions and should include the provisioning ofeconomic evidence of the scalability (or lackthereof) of human-centric methods for under-standing individual behaviour of customers.

Industry-Oriented Content

(J) Create industry-specific multi-channel cus-tomer journeys for key services industries suchas banking, insurance, retail and telecommu-nications. Link to and support these customerjourneys with knowledge-based representa-tions that bridge process and knowledge man-agement. This is a significant area of workthat will pave new integration of Process withKnowledge Management by creating a cus-tomer-centric knowledge-based organisation of


the enterprise. The meaning of the latter state-ment is about making all pertinent informa-tion from an enterprise to be organised and bemade available to customers in new, intelligentways in which "process footprints" serve as ahistorical base to reorganise and find informa-tion personalised to individual customers (thiscomment comes from a private communica-tion with P. Nandi).

Tooling

(K) Propose new tools that further the currentstate-of-the-art of Information Technology forprocess design and construction in the concertof a Business Process Engineering approach(in this connection, the generation of code is asecondary concern but flexible and open endto-end integrated capabilities would be a break-through). These process tools will be the car-rier of data analytics in real-time while sup-porting the delivery of personalised servicesto individual customers.

6 ConclusionsBusiness Process has left the productivity cornerwhere it has been confined by "scientific manage-ment". With the advent of customer-enterpriseinteractions of all forms and exercised throughmultiple channels, the need for a significantlyimproved customer experience is an imperativein transforming front-office operations. Conven-tional approaches to process have proven to havea devastating effect on loyalty. Renewed researchand professional efforts to approach process aspart of complex social systems are a must to copeFig. 4. Silos in information and process manage-ment have deepened with the evolution of eachdomain. This gap is more notorious after the ad-vent of business analytics, scorecards, perform-ance management and value-driven process BPMwith the challenges faced in those competences ofenterprises dealing with customers, particularlyin the business-toconsumer industries. BusinessProcess Engineering is a new domain of workthat attempts to make the past IT-centric viewof process into a multidisciplinary area of bothinstitution and practice knowledge.

7 Acknowledgements

The author would like to express his gratitudeto many colleagues for inspiring discussions onbusiness process in the context of Enterprise En-gineering, Business Informatics and practical con-notations of all these domains. A few names arementioned here: J. Verelst, H, Mannaert, J. Di-etz, E. Proper, A. Albani, P. Nandi, M. Indulska, J.Sapperstein, J. Sphorer, H. Hastings, M. Cefkin,J. Barjis, B. Hofreiter, C. Huemer, J. Tribolet, V.Becker, L. Flores and many others. Sincere thanksare also due to two referees for their invaluablecomments.

References

van der Aalst W. M. P., Berens P. J. S. (2001) Bey-ond Workflow Management: Product-drivenCase Handling. In: Proceedings of the 2001 In-ternational ACM SIGGROUP Conference onSupporting Group Work. GROUP ’01. ACM,Boulder, Colorado, USA, pp. 42–51 http://doi.acm.org/10.1145/500286.500296

van der Aalst W. M. (2004) Business ProcessManagement Demystified: A Tutorial onModels, Systems and Standards for WorkflowManagement. In: Desel J., Reisig W., Rozen-berg G. (eds.) Lectures on Concurrency andPetri Nets. Lecture Notes in Computer Sci-ence Vol. 3098. Springer, Berlin, Heidelberg,pp. 1–65 http://dx.doi.org/10.1007/978-3-540-27755-2_1

van der Aalst W. M. (2012) A Decade of Busi-ness Process Management Conferences: Per-sonal Reflections on a Developing Discipline.In: Barros A., Gal A., Kindler E. (eds.) Busi-ness Process Management. Lecture Notes inComputer Science Vol. 7481. Springer, Berlin,Heidelberg, pp. 1–16 http://dx.doi.org/10.1007/978-3-642-32885-5_1

van der Aalst W. M., van Hee K. M. (2002) Work-flowManagement: Models, Methods, and Sys-tems. MIT Press, Cambridge, MA, USA

http://doi.acm.org/10.1145/500286.500296

http://doi.acm.org/10.1145/500286.500296

http://dx.doi.org/10.1007/978-3-540-27755-2_1

http://dx.doi.org/10.1007/978-3-540-27755-2_1

http://dx.doi.org/10.1007/978-3-642-32885-5_1

http://dx.doi.org/10.1007/978-3-642-32885-5_1


66 Jorge L. Sanz

van der Aalst W. M., ter Hofstede A.,Kiepuszewski B., Barros A. (2003) WorkflowPatterns English. In: Distributed and ParallelDatabases 14(1), pp. 5–51 http://dx.doi.org/10.1023/A%3A1022883727209

van der Aalst W. M., Weske M., Grünbauer D.(2005) Case handling: a new paradigm forbusiness process support. In: Data & Know-ledge Engineering 53(2), pp. 129–162

Aguilar-Savén R. S. (2004) Business process mod-elling: Review and framework. In: Interna-tional Journal of Production Economics 90(2)Production Planning and Control, pp. 129 –149 http://www.sciencedirect.com/science/article/pii/S0925527303001026

Albani A., Dietz J. L. (2011) Enterprise ontologybased development of information systems.In: International Journal of Internet and En-terprise Management 7(1), pp. 41–63

Andjelkovic-Pesic M. (2007) Six Sigma philo-sophy and resource-based theory of compet-itiveness: An integrative approach. In: Factauniversitatis-series: Economics and Organiz-ation 4(2), pp. 199–208

Andjelković Pešić M. (2004) Six Sigma projectimplementation. In: YUPMA – VIII Interna-tional Symposium in Project Management,pp. 191–197

Andjelković Pešić M. (2006) Enterprises of NewEconomy – High Performance Organizations.In: Business Politics, pp. 54–58

Aveiro D., Silva A. R., Tribolet J. M. (2011) To-wards a G.O.D. theory for Organisational En-gineering: modelling the (re)Generation, Op-eration and Discontinuation of the enterprise.In: International Journal of Internet and En-terprise Management 7(1), pp. 64–83

Albani A., Barjis J., Dietz J. (eds.) Advancesin Enterprise Engineering III. Lecture Notesin Business Information Processing Vol. 34.Springer

Becker M. C. (2004) Organizational routines: areview of the literature. In: Industrial andCorporate Change 13(4), pp. 643–678

Bhattacharya K., Hull R., Su J. (2009) A data-centric design methodology for business pro-

cesses. In: Cardoso J., van der Aalst W. (eds.)Handbook of Research on Business ProcessModeling. IGI Global Snippet, Hershey, PA,USA, pp. 503–531

Bliss J. (2006) Chief Customer Officer: GettingPast Lip Service to Passionate Action. JohnWiley & Sons

Brumagim A. L. (1994) A hierarchy of corporateresources. In: Advances in strategic manage-ment 10(Part A), pp. 81–112

Capgemini (2012) World Retail Banking ReportCenter) A. A. P. Q. Process Classification Frame-

work (PCF). www.apqc.orgCohn D., Hull R. (2009) Business Artifacts: A

Data-centric Approach to Modeling BusinessOperations and Processes. In: IEEE Data Eng.Bull. 32(3), pp. 3–9

Davenport (1993) Process Innovation: Reengin-eering Work Through Information Techno-logy. Harvard Business School Press, Boston,MA, USA

Davenport T. H. (2005) The coming commod-itization of processes. In: Harvard BusinessReview 83(6), pp. 100–108

De Man H. (2009) Case Management: A Re-view of Modeling Approaches. http://www.bptrends.com

Debevoise T. (2007) Business Process Manage-ment with a Business Rules Approach: Imple-menting the Service Oriented Architecture.Tipping Point Solutions

Dietz J. L., Hoogervorst J. A., Albani A., AveiroD., Babkin E., Barjis J., Caetano A., HuysmansP., Iijima J., Kervel S. J. V. et al. (2013) Thediscipline of enterprise engineering. In: In-ternational Journal of Organisational Designand Engineering 3(1), pp. 86–114

Dietz J. (2011) A Business Process is More Thana Workflow. Tutorial at KEOD, Paris

Dumas M., Rosa M., Mendling J., Reijers H.(2013) Fundamentals of Business ProcessManagement. Springer

Eid-Sabbagh R.-H., Dijkman R., Weske M. (2012)Business Process Architecture: Use and Cor-rectness. In: Barros A., Gal A., Kindler E.(eds.) Business Process Management. Lec-

http://dx.doi.org/10.1023/A%3A1022883727209

http://dx.doi.org/10.1023/A%3A1022883727209

http://www.sciencedirect.com/science/article/pii/S0925527303001026

http://www.sciencedirect.com/science/article/pii/S0925527303001026

www.apqc.org

http://www.bptrends.com



ture Notes in Computer Science Vol. 7481.Springer, Berlin, Heidelberg, pp. 65–81 http://dx.doi.org/10.1007/978-3-642-32885-5_5

Felin T., Foss N. (2004) Organizational Routines:A Sceptical Look. DRUID Working Paper 04-13. Danish Research Unit for Industrial Dy-namics (DRUID)

Fleischmann A. (2011) Do We Need to Re-think Current BPM Research Issues? In:Fleischmann A., Schmidt W., Singer R., SeeseD. (eds.) Subject-Oriented Business ProcessManagement. Communications in Computerand Information Science Vol. 138. Springer,Berlin, Heidelberg, pp. 216–219

Franz P., Kirchmer M. (2012) Value-Driven Busi-ness Process Management: The Value-Switchfor Lasting Competitive Advantage. McGraw-Hill Education http://books.google.at/books?id=DYrOiwThP4UC

Harrison-Broninski K. (2005) Human Interac-tions: The Heart and Soul of Business ProcessManagement: how People Really Work andhow They Can be Helpful to Work Better.Meghan-Kiffer, Tampa, FL, USA

Hastings H., Saperstein J. (2007) Improve YourMarketing to Grow Your Business: Insightsand Innovation That Drive Business andBrand Growth, First. Wharton School Pub-lishing

Houy C., Fettke P., Loos P., Aalst W.,Krogstie J. (2010) BPM-in-the-Large To-wards a Higher Level of Abstraction in Busi-ness Process Management. In: Janssen M.,Lamersdorf W., Pries-Heje J., Rosemann M.(eds.) E-Government, E-Services and GlobalProcesses. IFIP Advances in Informationand Communication Technology Vol. 334.Springer, Berlin, Heidelberg, pp. 233–244http://dx.doi.org/10.1007/978-3-642-15346-4_19

Indulska M., Recker J., Rosemann M., GreenP. (2009) Business Process Modeling: Cur-rent Issues and Future Challenges. In:Eck P., Gordijn J., Wieringa R. (eds.) Ad-vanced Information Systems Engineering.Lecture Notes in Computer Science Vol. 5565.

Springer, Berlin, Heidelberg, pp. 501–514http://dx.doi.org/10.1007/978-3-642-02144-2_39

Khoyi D. (2010) Data Orientation. In: Swen-son K. D. (ed.) Mastering the Unpredictable.Meghan-Kiffer, Tampa, FL, USA

Klein M., Petti C. (2006) A handbook-basedmethodology for redesigning business pro-cesses. In: Knowledge and Process Manage-ment 13(2), pp. 108–119 http://dx.doi.org/10.1002/kpm.248

Ko R., Lee S., Lee E. (2009) Business processmanagement (BPM) standards: a survey. In:Business Process Management Journal 15(5),pp. 744–791

Krogstie J., Sindre G., Jørgensen H. (2006)Process models representing knowledge foraction: a revised quality framework. In:European Journal of Information Systems15(1), pp. 91–102

Le Clair C. (2012) Stuck In Cement: When Pack-aged Apps Create Barriers To Innovation.http://blogs.forrester.com

Levin D. Z. (2002) When Do OrganizationalRoutines Work Well? A New Approach toKnowledge Management. Rutgers University,Newark, USA

Lu R., Sadiq S. (2007) A Survey of Comparat-ive Business Process Modeling Approaches.In: Abramowicz W. (ed.) Business Informa-tion Systems. Lecture Notes in Computer Sci-ence Vol. 4439. Springer, Berlin, Heidelberg,pp. 82–94 http://dx.doi.org/10.1007/978-3-540-72035-5_7

Mannaert H., Verelst J. (2009) Normalized sys-tems: re-creating information technologybased on laws for software evolvability.Koppa Digitale Media BVBA, Kermt, Belgium

Miers D. (2009) Process Architecture: The Key toEffective Process Relationships. Presentationfrom BPM Focus

Nandi P. (2010) Introducing Business Entitiesand the Business Entity Definition Language(BEDL): A first class representation of datafor BPM applications. http://www.ibm.com

Nandi P., Sanz J. (2013) Cross-Functional Op-

http://dx.doi.org/10.1007/978-3-642-32885-5_5

http://dx.doi.org/10.1007/978-3-642-32885-5_5

http://books.google.at/books?id=DYrOiwThP4UC

http://books.google.at/books?id=DYrOiwThP4UC

http://dx.doi.org/10.1007/978-3-642-15346-4_19

http://dx.doi.org/10.1007/978-3-642-15346-4_19

http://dx.doi.org/10.1007/978-3-642-02144-2_39

http://dx.doi.org/10.1007/978-3-642-02144-2_39

http://dx.doi.org/10.1002/kpm.248

http://dx.doi.org/10.1002/kpm.248

http://dx.doi.org/10.1007/978-3-540-72035-5_7

http://dx.doi.org/10.1007/978-3-540-72035-5_7


68 Jorge L. Sanz

erations Modeling as a Nexus of Commit-ments: A New Approach for Improving Busi-ness Performance and Value-Creation. In:IEEE 15th Conference on Business Inform-atics (CBI 2013). IEEE, pp. 234–241

Näslund D. (2008) Lean, six sigma and leansigma: fads or real process improvementmethods? In: Business Process ManagementJournal 14(3), pp. 269–287

Nigam A., Caswell N. S. (2003) Business artifacts:An approach to operational specification. In:IBM Systems Journal 42(3), pp. 428–445

van Nuffel D. (2011) Towards designing modularand evolvable business processes. PhD thesis,Universiteit Antwerpen

Olbrich T. J. (2011) Why We Need to Re-think Current BPM Research Issues. In:Fleischmann A., Schmidt W., Singer R., SeeseD. (eds.) Subject-Oriented Business ProcessManagement. Communications in Computerand Information Science Vol. 138. Springer,Berlin, Heidelberg, pp. 209–215 http://dx.doi.org/10.1007/978-3-642-23135-3_12

Ould (1995a) Business Processes: Modelling andAnalysis for Re-Engineering and Improve-ment. John Wiley & Sons

Ould (1995b) Business processes: modelling andanalysis for re-engineering and improvement.Wiley, p. 224

Ould M. A. (1997) Designing a re-engineeringproof process architecture. In: Business Pro-cess Management Journal 3(3), pp. 232–247

Penrose E. (2009) The Theory of the Growth ofthe Firm. Oxford University Press

Pentland B. T., Feldman M. S. (2008) Designingroutines: On the folly of designing artifacts,while hoping for patterns of action. In: In-formation and Organization 18(4), pp. 235–250

Pentland B. T., Feldman M. S., Becker M. C., LiuP. (2012) Dynamics of organizational routines:a generative model. In: Journal of Manage-ment Studies 49(8), pp. 1484–1508

Proper H., Aveiro D., Gaaloul K. (2013) Ad-vances in Enterprise Engineering VII: ThirdEnterprise Engineering Working Conference,

EEWC2013, Luxembourg, May 13-14, 2013,Proceedings. Lecture Notes in Business In-formation Processing Vol. 146. Springer, Ber-lin, Heidelberg

Propp V. (1968) Morphology of the Folktale ScottL., Wagner L. (eds.), Second. University ofTexas Press, p. 184

Rawson A., Duncan E., Jones C. (2013) The truthabout customer experience. In: Harvard Busi-ness Review 91(9), pp. 90–98

Recker J., Rosemann M., Indulska M., Green P.(2009) Business Process Modeling - A Com-parative Analysis.. In: Journal of the Associ-ation for Information Systems 10(4)

Reijers H. A., Wijk S., Mutschler B., Leurs M.(2010) BPM in Practice: Who Is Doing What?In: Hull R., Mendling J., Tai S. (eds.) Busi-ness Process Management. Lecture Notes inComputer Science Vol. 6336. Springer, Berlin,Heidelberg, pp. 45–60 http://dx.doi.org/10.1007/978-3-642-15618-2_6

Rescher N. (1996) Process Metaphysics: An Intro-duction to Process Philosophy. SUNY seriesin philosophy. State University of New YorkPress

Robinson K. (1979) An entity/event data model-ling method. In: The Computer Journal 22(3),pp. 270–281

Rosenberg A., von Rosing M., Chase G., OmarR., Taylor J. (2011) Applying Real-World BPMin an SAP Environment, 1st. SAP Press

Rosenquist C. (1982) Entity Life Cycle Modelsand their Applicability to Information Sys-tems Development Life Cycles: A Frameworkfor Information Systems Design and Imple-mentation. In: The Computer Journal 25(3),pp. 307–315

Sanz J. L. (2011) Entity-centric operations mod-eling for business process management - Amultidisciplinary review of the state-of-the-art. In: IEEE 6th International Symposium onService Oriented System Engineering (SOSE2011). IEEE, pp. 152–163

Sanz J. L., Leung Y., Terrizzano I., Becker V.,Glissmann S., Kramer J., Ren G.-J. (2012)Industry Operations Architecture for Busi-

http://dx.doi.org/10.1007/978-3-642-23135-3_12

http://dx.doi.org/10.1007/978-3-642-23135-3_12

http://dx.doi.org/10.1007/978-3-642-15618-2_6

http://dx.doi.org/10.1007/978-3-642-15618-2_6


ness Process Model Collections. In: Daniel F.,Barkaoui K., Dustdar S. (eds.) Business Pro-cess Management Workshops. Lecture Notesin Business Information Processing Vol. 100.Springer, Berlin, Heidelberg, pp. 62–74 http://dx.doi.org/10.1007/978-3-642-28115-0_7

Schönthaler F., Vossen G., Oberweis A., KarleT. (2012) Business Processes for BusinessCommunities: Modeling Languages, Methods,Tools. Springer

Sharp A., McDermott P. (2009) Workflow Model-ing: Tools for Process Improvement and Ap-plications Development, First. Artech House,Inc., Norwood, MA, USA

Sidorova A., Isik O. (2010) Business processresearch: a cross-disciplinary review. In:Business Process Management Journal 16(4),pp. 566–597

Siegel J. (2008) In OMG’s OCEB CertificationProgram, What is the Definition of BusinessProcess? Certification Whitepaper, OMG

Singer R., Zinser E. (2011) Business ProcessManagement Do We Need a New ResearchAgenda? In: Fleischmann A., Schmidt W.,Singer R., Seese D. (eds.) Subject-OrientedBusiness Process Management. Communic-ations in Computer and Information Sci-ence Vol. 138. Springer, Berlin, Heidelberg,pp. 220–226 http://dx.doi.org/10.1007/978-3-642-23135-3_14

Smith H., Fingar P. (2007) Business Process Man-agement: The Third Wave. Meghan-KifferPress

Stone D., Devine J. (2013) From Moments toJourneys: A Paradigm Shift in Customer Ex-perience Excellence

Swenson K. D. et al. (2010) Mastering the Unpre-dictable: How Adaptive Case ManagementWill Revolutionize the Way that KnowledgeWorkers Get Things Done. Meghan-Kiffer,Tampa, FL, USA

Toussaint P. J., R Bakker A., Groenewegen L. P.(1998) Constructing an enterprise viewpoint:evaluation of four business modelling tech-niques. In: Computer methods and programsin biomedicine 55(1), pp. 11–30

Trkman P. (2010) The critical success factors ofbusiness process management. In: Interna-tional Journal of Information Management30(2), pp. 125–134

Tsoukas H. (2001) What Does it Mean to be Sens-itive to Process? Keynote address EuropeanGroup Organization Studies Colloquium 17,Lyon, France

Tsoukas H., Chia R. (2002) On organizationalbecoming: Rethinking organizational change.In: Organization Science 13(5), pp. 567–582

Van de Ven A. H. (1992) Suggestions for study-ing strategy process: a research note. In: Stra-tegic management journal 13(5), pp. 169–188

Villanueva J., Hanssens D. M. (2007) Cus-tomer equity: Measurement, managementand research opportunities. Foundations andTrends in Marketing 1 Vol. 1. Now PublishersInc, Hanover, MA, USA, pp. 1–95

Weske M. (2012) Business Process Management:Concepts, Languages, Architectures, Second.Springer

White S. (2004) Process modeling notations andworkflow patterns. http : / /www.bptrends .com

Yan Z., Dijkman R., Grefen P. (2012) Businessprocess model repositories–Framework andsurvey. In: Information and Software Techno-logy 54(4), pp. 380–395

Jorge L. Sanz

Business Analytics CenterSchool of Computing – Business SchoolNational University of SingaporeSingapore

http://dx.doi.org/10.1007/978-3-642-28115-0_7

http://dx.doi.org/10.1007/978-3-642-28115-0_7

http://dx.doi.org/10.1007/978-3-642-23135-3_14

http://dx.doi.org/10.1007/978-3-642-23135-3_14




70 Eng K. Chew

Eng K. Chew

Service Innovation for the Digital World

The foundational principles and conceptual building blocks of customer-centric service innovation (SI) practiceare explained, and a resultant integrated framework of SI design practices for customer value co-creationis synthesised. The nexus of service strategy, service concept and business model is identified to assure SIcommercialisation. The requisite SI models and processes to systematise the innovation practice are reviewed.The emergent practices of customer and community participation, in a digital world, across the firm’s entireSI lifecycle are explicated, together with the requisite strategic management practices for successful serviceinnovation.

1 Introduction

Service innovation – the art and science of cre-ating innovative services that customers valueand are willing to pay for – in the digital worldexemplifies many of the fundamental challengesof business informatics. Recent studies of ser-vice innovation have focused on the effectivemanagement of service innovation to enhancefirm performance – such as the importance ofmanaging inter-organisational relationships andcommitments (Eisingerich et al. 2009), the ante-cedents and consequences of service innovation(Ordanini and Parasuraman 2011), and a prelim-inary service-thinking framework for value cre-ation (Hastings and Saperstein 2013). These stud-ies have shown that success in service innova-tion requires "service thinking" (and attendantservice culture) and is contingent on effectivecollaboration with the firm’s customers and part-ners in the overall innovation process. Theseauthors also concur that service innovation isabout the creation of customer value (Grawe etal. 2009). However, the art and science of design-ing and managing service innovations, especiallyfor the digital world, remains an under-exploredresearch area. This paper seeks to contributeto filling this void by exploring the extant liter-ature to identify the critical constitutive theor-ies and practices that would lead to successfulservice innovation in line with Eisingerich et al.

(2009); Hastings and Saperstein (2013); Ordaniniand Parasuraman (2011). It focuses, in particu-lar, on the various critical roles of customers invalue co-creation for themselves in conjunctionwith the service provider and their network ofpartners.

The paper is structured as follows. Section 2 de-scribes in detail the fundamental building blocksof service innovation: service dominant logic, ser-vice systems, operant resources and dynamic cap-abilities, value networks, and finally, customervalue co-creation – the ultimate purpose of ser-vice innovation. Section 3 synthesises from theextant literature a framework of design practicesfor service innovation, comprising four businessstrategy-aligned interrelated practices of serviceconceptualisation, service design, customer ex-perience and value creation, and service architec-ture which, collectively, are typically pursued bydesigners iteratively (experimentally) and hol-istically. Section 4 links the design practicesframework to service strategy on one hand andbusiness model design on the other to addressthe commercialisation aspect of service innova-tions. Section 5 reviews, individually, the com-mon and foundational service innovation func-tional models (in terms of the ’scope’ of and the’competence-based approach’ to service innov-ation) and processes (in terms of new servicedevelopment) for the creation of customer value.

Enterprise Modelling and Information Systems ArchitecturesVol. 9, No. 1, June 2014Service Innovation for the Digital World 71

Section 6 and Sect. 7, respectively, review the in-creasingly important ’open innovation’ practicesof involving customers and online communityin the end-to-end service innovation process inthe digital world, while Sect. 8 addresses the re-quisite strategic management capability to en-sure service innovation success. Finally, Sect. 9concludes the paper by summarising the requis-ite principles (theories) and service design andinnovation management practices for service in-novation excellence.

2 Conceptual Building Blocks

2.1 Service Dominant Logic

In the early days (pre-1980) of services marketing,services were seen as a special kind of products.Seen as a unit of production output, services weredefined as residues of, and thus subordinate to,products (Lovelock and Gummesson 2004; Vargoand Lusch 2004). From this goods production per-spective, services as an output are seen to possessfour so-called IHIP characteristics which are dis-tinctly different from physical products: Intangib-ility, Heterogeneity, Inseparability and Perishabil-ity (Lovelock and Gummesson 2004). Intangibilityrefers to the services output as being intangible.Heterogeneity refers to the services possessingvariable input resources and performance out-comes. Inseparability refers to the productionand consumption of services occurring simultan-eously. Perishability refers to the services outputas being non-durable and non-storable. How-ever, these services characteristics were actuallyshown to be not generally applicable to all ser-vices (Lovelock and Gummesson 2004). Leadingservice scholars around the globe also regardthe production-oriented IHIP view as outdated(Edvardsson et al. 2005), because it fails to cap-ture the processual, interactive and relationalnature of service co-creation and consumption asseen from the customer perspective (Edvardssonet al. 2005; Fitzsimmons and Fitzsimmons 2010).This alternative customer-centric and relational

view constitutes the service-dominant logic (S-DL) which defines service as a process of apply-ing the competencies and skills of a provider forthe benefit of, and in conjunction with, the cus-tomer (Vargo and Lusch 2004, 2008). A serviceoffering is produced using the firm’s resourcesincluding both tangible (such as goods) and in-tangible (such as knowledge, competence andrelationship) assets (Arnould 2008). The valuecharacteristics of the service provisioned, how-ever, are co-created through the interactions ofthe client’s competences with that of the serviceprovider (Gallouj 2002). Thus the client is act-ive in a service interaction; it co-creates value(for itself) with the provider (Fitzsimmons andFitzsimmons 2010; Gadrey and Gallouj 2002; Gal-louj 2002). The central idea of S-D logic is that"exchange is about the process of parties doingthings for and with each other, rather than trad-ing units of output, tangible or intangible" (Vargoand Lusch 2008).

2.2 Service Systems

Service systems are the basic unit of analysis of(the customer-centric view of) service (Maglioand Spohrer 2008). A service system is defined asa complex adaptive system of people, and tech-nologies working together to create value for itsconstituents (Spohrer et al. 2007). For example,a telecom company is a complex market-facingtechnology-based service system. The study ofservice systems is focused on creating a basisfor systematic service innovation (University ofCambridge and IBM 2007). It requires a mul-tidisciplinary integrative understanding of theways organisation, human, business and tech-nology resources and shared information maybe combined to create different types of servicesystems; and how the service systems may inter-act and evolve to co-create value (Maglio et al.2009). A service system has a service providerand a service client or beneficiary (Maglio et al.2006). Service systems are connected by valuepropositions (Maglio et al. 2009). IT or business


72 Eng K. Chew

process outsourcing service configurations nego-tiated and agreed to between service providersand clients are examples of service systems. Con-sistent with S-DL, value-cocreation interactionsbetween service systems are service interactions,each comprising three main activities: propos-ing a value-cocreation interaction to another ser-vice system (proposal); agreeing to the proposal(agreement); and realising the proposal (realisa-tion) (Maglio et al. 2009).

Service systems are dynamic, constantly compos-ing, recomposing and decomposing over time. Aservice system operates as an open system cap-able of improving the state of another systemthrough sharing or applying its resources (in-cluding competences/capabilities), and improv-ing its own state by acquiring external resources(Maglio et al. 2009). Thus, service systems engagein knowledge-based interactions to co-createvalue, where value is derived and determinedin use – the integration and application of re-sources in a specific context embedded in firm’soutput – and captured by price (Vargo et al. 2008).Consequently, advances in service innovation areonly possible when a service system has inform-ation about the capabilities and the needs of itsclients, its competitors and itself (Maglio et al.2009).

Integral to and as a consequence of service innov-ation, service systems co-create value throughcollaboration and adaptation, and establish a bal-anced and interdependent framework for sys-tems of reciprocal service provision. Servicesystems survive, adapt, and evolve through ex-change and application of resources (especiallyknowledge and skills -operant resources as ex-plained below) with other systems (Vargo et al.2008).

2.3 Operant Resources & DynamicCapabilities

A resource is called an operand resource (i.e.,tangible physical resource) "on which an opera-tion or act is performed to produce an effect", or

an operant resource (i.e., intangible knowledge-based capability) "which acts on other operandor operant resource to produce an effect" (Vargoand Lusch 2004). Operant resources are dynamic,which include competences or capabilities thatcan be nurtured and grown in some unique waysto provide competitive advantage to firms(Madhavaram and Hunt 2008). Operant resourcesthat are valuable, rare, inimitable and not sub-stitutable will generate sustainable competitiveadvantage for firms. For example, market orient-ation – i.e., market sensing and customer linkingcapabilities – is an operant resource that wouldcreate that advantage (Arnould 2008). This mo-tivates firms to create and use dynamic operantresources to sustain the competitive advantage.

Highly innovative firms possess "masterfully de-veloped" operant resources accumulated over along period from institutionalised learning prac-tices (Madhavaram and Hunt 2008). These re-sources allow the firm to effectively manage co-evolution of knowledge, capabilities, and productsor services to sustain its competitive advantage.Collaborative competence is identified as a pivotaloperant resource for sustained service innovation(Lusch et al. 2007) – one that assists in the devel-opment of two additional meta-competences: ab-sorptive competence, and adaptive competence(also collectively known as dynamic capabilit-ies (Teece 2007)) which enable the firm to, re-spectively, absorb new knowledge and inform-ation from partners, and adapt to the complexand turbulent environments by reconfiguring itsresources (and organisational capabilities) withthose of the external partners. These operant re-sources are key components of a service systemwhich is conceptualised as a resource integrator(Spohrer et al. 2007). It is the people’s uniqueknowledge and skills and dynamic capabilitiesthat make service systems adaptive to and sus-tainable with the changing market environments(Spohrer et al. 2007; Teece 2007; Vargo et al. 2008).

2.4 Value Networks of Digital WorldIn the increasingly digital world, informationtechnologies are "liquefying" physical assets into


information resources, and transform a servicefirm into a value-creating service system in whicha constellation of economic actors (customers,suppliers, business partners and the like) are ableto seamlessly collaborate to co-create value (Nor-mann and Ramirez 1993). This reflects the S-Dlogic’s commitment to explicating the firm’s col-laborative processes with customers, partners,and employees to engage in the co-creation ofvalue through reciprocal service provision (Luschet al. 2007). And the customer is regarded as anoperant resource – a dynamic resource that iscapable of acting on other resources to createvalue for itself (Vargo and Lusch 2008).

With the ubiquitous digitalisation, goods are in-creasingly being embedded with microprocessorsand intelligence and becoming versatile platformsfor service provision with enhanced customerand supplier insights and superior self-serviceability. It also reduces transport and commu-nications costs, enhances the ability to interactdirectly with customers and suppliers and con-sequently coordination between firms becomesmore efficient and responsive (Lusch et al. 2009).Thus, the firm will become an essential serviceprovisioning agent (actor) in a complex and ad-aptive value network comprising a spatial andtemporal structure of loosely coupled value-pro-posing social and economic actors. The actorsinteract through institutions and technology cap-able of spontaneously sensing and respondingvia their dynamic capabilities to co-produce ser-vice offerings, exchange service offerings, andfinally co-create value. They are linked togetherby means of competences, relationships, and in-formation (Lusch et al. 2009). The relationshipsare collaborative and guided by non-coercive gov-ernance. This implies voluntary, reciprocal useof resources for mutual value creation by two ormore interacting actors, through the symmetricexchange of information and resources (compet-ences) (Vargo et al. 2008). So in the value network,customers and suppliers become partners, andcompetitors become collaborators as well (Ches-brough and Davies 2010). Each firm (actor) oper-

ates as an open system (Maglio et al. 2009). Firmsmust practice open innovation (Chesbrough 2003)and develop systems integration capability (Ches-brough and Davies 2010) as part of its dynamiccapabilities (Teece 2007) to integrate the requisitecompetences and resources from external sourceswith their own to co-create value; e.g., Apple’screation of the iPod/iTune music service.

Value co-creation and innovation in the digitalworld would require firms to institute individu-alised and immediate customer feedback (to andfrom the customers) to engender customer andorganisational learning (Johannessen and Olsen2010). This requires a new IT-enabled organisa-tional logic which encompasses modular (multi-sourcing) flexibility, front-line (customer learn-ing) focus, IT-enabled individualisation and "con-nect and develop" innovation practices (Ches-brough and Davies 2010; Johannessen and Olsen2010). In addition, the firm needs new cooper-ation structures by partaking in global compet-ence clusters and practising coopetition (Johan-nessen and Olsen 2010). Above all, to be agileand adaptable as they learn of changing customerneeds, firms need to develop dynamic operantresources – the dynamic capabilities (Teece 2007).The dynamic capabilities allow firms to continu-ally align their competences to create, build andmaintain relationships with (thus the value pro-positions to) customers (the ultimate source ofrevenue) and suppliers (the source of resourceinputs).

2.5 Customer Value Co-creation

Customer is at the heart of value creation and ser-vice is about relationship with the customer (Ed-vardsson et al. 2005). The customer interacts withthe service provider via the interface throughwhich information /knowledge, emotions and ci-vilities are exchanged to co-create value (Gallouj2002). Value is wholly determined by the cus-tomer upon, and in the context of, service usage(and customer experience), in which the compet-ence (operant resource) of the provider is integ-rated with the competence (operant resource) of


74 Eng K. Chew

the customer to (perform ’a job’ to) create (busi-ness) value with the customer (Edvardsson et al.2005; Vargo and Lusch 2008). The service pro-vider cannot deliver value, but only offer valuepropositions (Vargo and Lusch 2008). To win theservice game, the value proposition must consist-ently meet the customer expectations and beha-vioural needs (Schneider and Bowen 2010). Thiscan be assured by co-opting the customer com-petence in co-creating the service offering withthe provider (Prahalad and Ramaswamy 2000) –e.g., user toolkits for innovation (Hippel 2001).However, the customer would collaborate withthe provider in co-creation of core service offer-ings only if they would gain benefits, such as:expertise, control, physical capital, risk taking,psychic benefits, and economic benefits (Luschet al. 2007).

Service innovation must therefore be concernedwith effectiveness of value co-creation betweenthe provider and beneficiary. It recognises theprinciple that a proposed value by the provider,in the context of the client, is actually a compos-ite of benefits and burdens (or costs), which canbe evaluated using a customer value equation(Fitzsimmons and Fitzsimmons 2010). Burdens re-late to the service’s usability (or its relative ease-of-integration with the client’s resources or activ-ities to "perform the job the service is hired todo") – the more user-friendly it is the less the bur-den and the greater the user experience; and thegreater the customer efficiency (Xue and Harker2002). Thus, the most compelling service withthe best "value for money" to the client is onethat has the largest "benefit-to-costs" ratio. Thissuggests that user involvement in co-creating theservice offerings (or co-designing the value pro-positions) with the provider would more likelycreate ’fit-for-purpose’ service for the client andthereby maximising the benefit.

Service firms must therefore "consider not onlythe employees’ productivity but also the ’pro-ductivity’ and experience of the customer" (Fitz-simmons and Fitzsimmons 2010; Lusch et al. 2007;Schneider and Bowen 2010; Womack and Jones

2005). From a service system viewpoint, value,created as a result of integrating the provider’sresources with the client’s, increases the clientsystem’s adaptability and survivability to fit withits changing environment (Vargo et al. 2008).

3 Framework of Design Practices

To create innovative services that sustainably co-create superior customer value in the constantlyevolving value networks of the digital world, adesign framework is synthesised from the ex-tant literature consistent with the preceding con-ceptual building blocks. The design frameworkfor service innovation consists of closely inter-related practices of: (a) service concept whichdefines what the service is and how it satisfiescustomer needs, (b) service design which definesthe service delivery mechanisms to consistentlysatisfy customer needs, (c) customer experienceand value creation which guides service designto align the provider’s competences and learn-ing regime to those of the customers to ensuresuperior experience, and (d) service architecturewhich systematises service design and innova-tion. These four interrelated practices are de-tailed below individually, but are typically prac-tised in the real-world iteratively and holistically.

3.1 Service Concept

A service concept defines the conceptual modelof the service. It describes what the service isand how it satisfies customer needs (Bettencourt2010). Service concept is the most critical com-ponent of service strategy, and reflects the align-ment of the customer needs (job and outcome op-portunities) with the company capabilities. Ser-vice concept also forms the fundamental partof service design, service development and ser-vice innovation (Fynes and Lally 2008). It is de-veloped as the end-result of the activities of stra-tegic positioning, idea generation and conceptdevelopment/refinement. The conceptual modelof a service consists of seven components whichtogether define the desired customer outcomes


(value propositions) of the service: service bene-fits, participation activities, emotional compon-ent, perception component, service process, phys-ical environment, and people/employee (Fynesand Lally 2008). To define an innovative serviceconcept, Bettencourt (Bettencourt 2010) recom-mends that a service firm should: focus creativeenergies on specific job and outcome opportunit-ies; identify where the key problems lie in satis-fying high-opportunity jobs and outcomes; sys-tematically consider a diverse set of new serviceideas to satisfy the opportunities; and build a de-tailed concept with service strategy and servicedelivery in mind.

Service concept is the principal driver of servicedesign decisions at all levels of planning and im-plementation. It relates to service architectureor service blueprint which guides service design,and to service governance which defines the de-cision rights and the decision making processfor service design, planning and implementation(Goldstein et al. 2002). For example, at the stra-tegic planning level, the service concept drivesdesign decision for new or redesigned services.At the operational level it defines how the servicedelivery system implements the service strategyand how to determine appropriate performancemeasures for evaluating service design. At theservice recovery level, it defines how to designand enhance service encounter interactions. Thusservice concept is the common foundation fornew service development, service design and ser-vice innovation. For instance, service conceptdevelopment and testing is at the heart of servicedesign in new service development. Central toservice conceptualisation is declaring what thecustomer value proposition is in relation to thefirm’s strategic intent, how it meets the customerneeds and what is the service logic required indelivering the value proposition (Goldstein et al.2002). Service concept articulates the service op-eration – why and how the service is delivered;the service experience – i.e., customer’s experi-ence; the service outcome – i.e., customer bene-fits; and the service value – i.e., the perceived

customer benefits minus the service cost (Clarket al. 2000). Service concept and the correspond-ing service design (described below) are intendedto reflect the service firm’s business strategy andtherefore directly impact the firm’s financial per-formance. From the perspective of service innov-ation (or new service development) process (de-tailed in Sect. 5.2) service concept is developed inthe "Create Ideas" phase and selected for designin the "Evaluate and Select Ideas" phase (after ex-perimentation), while the corresponding servicedesign is developed in the "Plan, Design Developand Implement Ideas" phase. However, in thedigital world, the innovation process would tendto be circularly iterative akin to "agile (emergent)development" as opposed to a purely linear (pre-dictive) manner.

3.2 Service Design

Service design starts with the customer/user anddefines how the service will be performed usinghuman-centred and user-participatory methodsto model the service performance (Holmlid andEvenson 2008). A service is conceptualised as anopen system with customers being present every-where. Service design must address strategicservice issues such as marketing positioning andthe preferred type of customer relationship, inline with the strategic intent of the service or-ganisation. Service governance is also requiredto monitor the service qualities and financial per-formance against the design outputs. The frame-work for designing the service delivery systemmust address multiple interrelated factors: stand-ardisation; transaction volume per time period;locus of profit control; types of operating person-nel; types of customer contacts; quality control;orientation of facilities; and motivational char-acteristics of management and operating person-nel (Goldstein et al. 2002). The service deliverysystem fulfills the firm’s strategic service visionand is designed/specified by means of serviceblueprinting (Bitner et al. 2008; Fitzsimmons andFitzsimmons 2010). Service blueprinting is a mapor flowchart of all the transactions constituting


76 Eng K. Chew

the service delivery process. The map identifies:the potential ’fail-points’; the line of interactionbetween client and provider known as service en-counters; the line of visibility – above it employ-ees actions are visible to the customer (directlyaffecting customer experience); below it is the’back-stage’ ; and the internal line of interactionsbelow the line of visibility (Bitner et al. 2008;Fitzsimmons and Fitzsimmons 2010). The serviceencounter design is a critical element of servicedesign, because from the customer’s viewpoint"these encounters ARE the service" (Bitner et al.2008). The design focuses on maximising the qual-ity of ’service experience’ by the customer. How-ever, service experience is the result of the com-bined efforts of the ’back stage’ information andprocesses and the ’front stage’ customer handling– both must work seamlessly in unison in satis-fying the customer request (Glushko and Tabas2009).

Taking an end-to-end view of service processallows designers to analyse the stakeholders’ re-quirements, pain points and performance met-rics from which service design (or redesign foran existing service) could be developed in col-laboration with the stakeholders incorporatinga combination of changes across process, organ-isation, technology, and tool in an integrativemanner (Maglio et al. 2006).

Service encounter design is guided by the pos-sible relationships between the three parties inthe service encounter: the service organisation(whether to pursue a service strategy of efficiency(cost leadership) or effective (customer satisfac-tion) or both); the contact personnel (followingstrict rules/order or empowered with autonomyand discretion); and the interaction between con-tact personnel and the customer (balancing con-flicting "perceived control" by both parties) (Fitz-simmons and Fitzsimmons 2010). Technologycould be designed into the service encounter infour ways: (a) technology-assisted service en-counter – only the contact personnel has access

to the technology; (b) technology-facilitated ser-vice encounter – both the customer and the con-tact personnel have access to the technology;(c) technology-mediated service encounter – thecustomer and contact personnel are not physic-ally co-located and their interaction is mediatedthrough the (online) technology; (d) technology-generated service encounter – i.e., self-service,the contact personnel is completely replaced bytechnology (Fitzsimmons and Fitzsimmons 2010;Froehle and Roth 2004). Thus technological innov-ation in services could require a change in cus-tomer role in the service delivery process. There-fore it is critical to take into account the potentialcustomer (as well as employee) reaction to thenew technology in the design phase to avoid fu-ture problems of acceptance (Fitzsimmons andFitzsimmons 2010).

Service design must include strategies for hand-ling service variability to ensure sustained levelof service quality expected by customers (Glushkoand Tabas 2009). For instance, to manage an unex-pected deviation from normal service encounter,the service design (per service strategy and gov-ernance) may incorporate the notion of servicepersonnel ’empowerment’ which grants themthe discretion to recover from service deviation(failure) by offering ’compensations’ or altern-ative solutions to the customer to minimise ad-verse impacts to the customer (Glushko and Ta-bas 2009). Moreover, where multichannel servicesare provided, the design must ensure consistentservice experience across all channels. Finally,service design needs to incorporate the require-ments of lean consumption (Womack and Jones2005) and achieve the objectives of service profitchain (Heskett et al. 2008).

Design of a service system (which offers the ser-vice) similarly must address the roles of people,technology, shared information, as well as therole of customer input in production processesand the application of competences to benefit oth-ers. The design must also address the service sys-tems’ requirements for agility and adaptabilityin alignment with their environments (Spohrer


et al. 2007). A learning framework is necessary tosustain the firm’s creative design ability, and im-prove and scale the service systems. The frame-work is designed to achieve three critical require-ments: effectiveness – the right things get done;efficiency – things are done in the right way;sustainability – the right relationships exist withother service systems to ensure the system’s longterm sustainability (Maglio et al. 2009; Spohreret al. 2007). Sustainability is achieved throughthe service system’s (brand) reputation, becauseexcellent reputations naturally attract value pro-positions from other service systems wantingto co-create value. It also requires appropriateamount of shared information to be available toall service systems (the principle of informationsymmetry) to enhance coordination and mutualsustainability within the service ecosystem. Thedesign is however inherently challenged by thepeople factor, as people are complex and adapt-ive.

In sum, service system design, broadly, must ad-dress four variables: physical setting; processdesign – the service blueprinting or mappingwhich designs ’quality’ into the service deliv-ery system; job design – the social technical jobdesign which include addressing the employeemotivational requirements; and people – the staff(competence) selection (Goldstein et al. 2002).

3.3 CustomerExperience & ValueCreation

Customer experience requirements of each ser-vice type are usually analysed using use-casescenarios similar to that of service blueprint (Bit-ner et al. 2008; Patricio et al. 2008). Customerexperience is influenced by the service intens-ity, which is defined in terms of the number ofactions initiated by the service provider, or theamount of information exchanged in a service en-counter or the duration of the service encounter(Glushko and Tabas 2009). The service design ofmulti-interface system must unify service man-agement, human computer interface, and soft-ware engineering perspectives into an integrated

design embodying the customer experience re-quirements (Patricio et al. 2008).

Service organisations are increasingly managingcustomer experiences to promote differentiationand customer loyalty. The experience-centric ser-vice providers design the activity and contextof the experience to engage customers in a per-sonal, memorable way. The experience designmust address the dynamic and ongoing engage-ment process between customers and the serviceorganisation. The engagement can be emotional,physical, intellectual, or even spiritual, depend-ing on the level of customer participation andthe connection with the environment (Zomerdijkand Voss 2010).

Customer value creation process is a dynamic, in-teractive, non-linear and often unconscious pro-cess (Payne et al. 2008). Value is in the contextof the performance outcome of the customer’sresource integration practice. To ensure optimalvalue co-creation, the three contiguous processes:the customer value-creating processes; the sup-plier value-creating processes and the interfacingservice encounter processes must all be aligned(Payne et al. 2008). The customer experience is aculmination of the customer’s cognitions, emo-tions and behaviour during the relationship withthe supplier. These elements are interdependentand involve the customer in thinking, feeling anddoing – leading to customer learning – in theprocess of value co-creation (Payne et al. 2008).Indeed, a recent study by (Helkkula et al. 2012)showed that "value in the [customer] experience[is characterised] as an ongoing, iterative circu-lar process of individual, and collective customersense making, as opposed to a linear, cognitiveprocess restricted to isolated service encounters."(p.59) More research is required on "the need forappropriate metrics for the cognitive and emo-tional demands" of customer experience imposedby different service interaction designs (Glushkoand Tabas 2009).


78 Eng K. Chew

3.4 Service Architecture

Service architecture is conceptualised to system-atise service design and innovation. Leveragingconcepts from product architecture, service ar-chitecture aims to create a common language(comprised of nodes and linkages) across differ-ent views on service design and a systematic wayto operationalise and measure the degree of ser-vice architecture modularity (Voss and Hsuan2009).

Service architecture is constituted in accordancewith the principle of modularity, which in turnis characterised by five dimensions: compon-ents and systems as the basic modular units, theinterfaces, degree of coupling, and commonal-ity sharing between components, and platformas the overarching configuration of componentsand interfaces that makes up the product/servicearchitecture (Fixson 2005). Modularity refers tothe degrees by which interfaces between com-ponents are standardised and specified to allowfor greater re-usability and sharing of (common)components among product/service families. Itprovides the basis for mixing and matching ofcomponents to meet the mass-customisation re-quirements; yields economies of scale and scope,and can help structure products/services to fa-cilitate outsourcing. Platform strategies are thevehicles for realisation of mass customisation(Fixson 2005). As platform decisions often cutacross several product/service lines or divisionalboundaries, platform strategic decisions mustbelong in the top management team who needto and can resolve cross-functional conflicts tojointly-achieve the firm overall strategy.

An important and challenging aspect of servicearchitecture is the interface. Interfaces in ser-vices can include people, information, and rulesgoverning the flow of information. Service in-terface can also include the flow of people. Ingeneral, an active role in service customisationwould be played by both the front-end employ-ees and the customers themselves. This wouldsuggest the service components need to be more

loosely coupled than product components (Rothand Menor 2003).

A service system can be analysed, for the pur-poses of service architecture, in terms of fourlevels of increasing details in specification: in-dustry level, service company/supply chain level,service bundle level, and service package/com-ponent level (Voss and Hsuan 2009). At level 0,the industry architectural template defines thevalue creation and the division of labour as wellas value appropriation and the division of surplusor revenue among the different players. At level1, the service company and its supply chain(s)are modelled both upstream and downstream.Both shared (internal cross-functional) and out-sourcing of service components are importantconsideration for the service company level foreconomic and resource flexibility reasons, in linewith its business strategy. At levels 2 and 3, theservice concept and service design activities ofservice innovation practice are harmonised andintegrated to assure service agility. At level 2,the individual service bundles of the service of-fering at the company level are analysed – eachbundle is viewed as a set of modules of servicedelivery, comprising the front- and back-officefunctions (and associated capabilities). At level 3,the service package and component level, thecharacteristics of the building blocks (compon-ents) are specified that contribute to the overallsystems architecture, namely: standardisation,uniqueness, degree of coupling and replicability(Voss and Hsuan 2009). Thus, service architectureenables service agility as new services can beprovisioned with minimal cost and little internalchange, and the architecture can be dynamic-ally adapted in response to external stimuli. Butthis would require support by a correspondingmodular organisational architecture as well as ISarchitecture (Voss and Hsuan 2009).

4 Service Strategy & Business ModelThere is a four-step approach to developing asuccessful service strategy: (1) Select the innova-tion focus, such as new service innovation or ser-vice delivery innovation, and the target customer


group(s); (2) Uncover customer needs in termsof jobs to get done and outcomes expected; (3)Prioritise customer needs; (4) Develop a servicestrategy (and attendant service concept) to fulfilthe high priority customer needs (Bettencourt2010). A successful service strategy fits what thecustomer will value with what the company candeliver. This means aligning the service concept(what it would take to deliver on the customervalue propositions), and hence service architec-ture, with firm’s capabilities, resources, cultureand strategy.

Experiences of leading companies, such as South-west Airlines, show that successful strategieswould include: (1) close coordination of the mar-keting and operations relationship; (2) a strategybuilt around elements of a strategic service vis-ion; (3) an ability to redirect the strategic serviceinward to focus on vital employee groups; (4) anappraisal of the effects of scale on both efficiencyand effectiveness; (5) the substitution of informa-tion for other assets; and (6) the exploitation ofinformation to generate new business (Heskett etal. 2008). In addition, six successful strategic prac-tices have been identified for service commer-cialisation: (1) leveraging fundamental sourcesof value that influence shareholder wealth, (2)managing customers’ perceptions of the servicevalue proposition, (3) creating an attractive fin-ancial architecture for customising pricing forprofitability, (4) ensuring service excellence in im-plementation, (5) planning for service recovery,and (6) managing the holistic service experience(including the servicescape) (Bolton et al. 2007).These successful strategic practices mirror thedesign of corresponding business model designconsiderations below and require superior collab-orative competence. This is because it leveragesthe firm’s dynamic capability to absorb informa-tion and knowledge from the environment, cus-tomers, and its value networks, and adapt theservice to respond to dynamic and complex en-vironments, while ensuring consistent superiorcustomer experience at each service encounterpoint.

Strategy defines the choice as to which businessmodel among many options to adopt for competi-tion in the marketplace. Thus the chosen businessmodel is a reflection of the service strategy – itrepresents the logic of the firm, the way it oper-ates and how it creates value for its stakehold-ers (Casadesus-Masanell and Ricart 2010; Oster-walder and Pigneur 2005). Service business modeldefines the end-to-end service delivery activities,in accordance with the service concept, by whichfirms deliver value to customers, entice custom-ers to pay for value, and convert those paymentsto profit (Osterwalder and Pigneur 2005; Teece2010). It articulates the logic, the data, and otherevidence that support a value proposition forthe customer, and a viable structure of reven-ues and costs for the enterprise delivering thatvalue. Business model embodies the organisa-tional and financial ’architectures’ of a business(Osterwalder and Pigneur 2005; Teece 2010). Abusiness model can be conceptualised as a sys-tem of interdependent (service delivery) activit-ies that transcends the focal firm and spans itsboundaries, and enables the firm, in concert withits partners, to create value and also to appropri-ate a share of that value. The service businessmodel is composed of two building blocks: (a)design elements – content, structure and gov-ernance that describe the architecture of a servicedelivery activity system (Level 2 and Level 3 ofservice architecture); (b) design themes – novelty,lock-in, complementarities and efficiency that de-scribe the sources of the service delivery activitysystem’s value creation (Zott and Amit 2010).

In sum, a service firm’s customer value propos-ition crystallised by the service concept servesas the bridge connecting its service strategy andbusiness model. The former defines the serviceconcept and service delivery mechanisms (con-sistent with the service architecture) while thelatter defines the revenue and cost models (finan-cial architecture) of the selected activity system(in accordance with the service delivery archi-tecture) designed to serve the targeted customersegments. Both practices tend to be pursued in


80 Eng K. Chew

parallel and interactively due to their close inter-relationship. And both practices are required tocreate and sustain the competitive advantage forthe firm.

5 Service Innovation Models andProcess

Service innovation is about the creation of cus-tomer value (Grawe et al. 2009). The source ofservice innovation opportunities is from discov-ering how customers define value – for instance,customers hire products and services or solutionsto get a job done; or use outcomes to evaluatesuccess in getting a job done; and have distinctneeds that arise related to the "consumption" of asolution (Bettencourt 2010). Four types of serviceinnovation can be identified from the customerviewpoint: (1) New service innovation – discov-ery of new or related jobs to get done; (2) Coreservice innovation – helping the customer get acore job done better; (3) Service delivery innov-ation – improving the ways a core job get done;(4) Supplementary service innovation – helpingthe customer get jobs done related to product us-age or consumption done (Fynes and Lally 2008).Service innovation can also be characterised bythe degree of interaction with the customer andthe degree of information asymmetry within theservice relationship (Gallouj 2002). This sectionreviews the common, foundational service innov-ation (functional and competence-based) modelsand processes for creating all types of innovativeservices that help customers get their jobs done.

5.1 Functional Model of ServiceInnovation

Service innovation is often a result of a combin-ation of conceptual, technological and organisa-tional innovations combined with new ways ofrelating to the consumer (Hertog 2002). A com-monly used functional model for identifying thefocus or vector of a service innovation consistsof four dimensions of service: (a) new service

concept – a new idea of concept of how to organ-ise a solution to a job/problem in a given mar-ket; (b) new client interface – new information-centric (often online) personalised interface (Gal-louj 2002) to facilitate service offering co-design,co-production and value co-creation with the cli-ents; (c) new service delivery system and organ-isation in line with the firm’s strategic servicevision and new service concept; and (d) techno-logy options – the specific role of technologyselected1 (Gallouj 2002) in the service innova-tion (Hertog 2002). Thus service innovation is amulti-dimensional phenomenon. A completelynew service (radical innovation) usually meansinnovations in all the above four dimensions.On the other hand, incremental service innov-ation means innovation in one or more of theabove four dimensions. Equally important isthe need to address the linkages between thesedimensions in order to implement the serviceinnovation, as they represent the requisite mar-keting, organisational development and learningprocesses (human resource) (Gallouj 2002; Maglioet al. 2009; Spohrer et al. 2007) and distribution(supply chain/logistics) capabilities to realise theinnovation. For example, launching a new ser-vice concept requires marketing expertise. Thedecision as to whether to develop new servicesrequires organisational knowledge: the organisa-tional capabilities required versus available andsuitability of existing organisational structure todeliver the service (Gallouj 2002; Hertog 2002).Thus while service innovation may arise fromchanging one of the above four dimensions, itrequires interdisciplinary collaboration betweenmarketing, human resource, distribution and ITto bring about the change and take the innova-tion to market. In sum, each particular (type of)service innovation is characterised by the com-bination of the four dimensions: the weight ofthe individual dimensions and the relative sig-nificance of the various linkages between them

1Use of technologies in service firms tends to followthe so-called "Barras reverse product cycle RPC" model –start with back-end then front-end process innovations andfinally product/service innovation (Gallouj 2002).


(Hertog 2002). To co-create and capture value forthe innovative firm, a new business model mustbe designed that reflects the operating and finan-cial model of the service concept and associatedlinkages to the other dimensions (Teece 2010).

5.2 Competence-based Model of ServiceInnovation

There are three different approaches to definingand studying service innovation (Gallouj 2002):an assimilation or technologist approach, whichtreats services as similar to manufacturing; a de-marcation or service-oriented approach, whichdistinguishes services (possessing the aforemen-tioned IHIP characteristics) from manufacturinginnovation; and a synthesis or integrative ap-proach, which suggests that service innovationbrings to the forefront hitherto neglected ele-ments of innovation that are of relevance formanufacturing as well as services. The synthesisor integrative approach is widely adopted andit is congruent to the service-dominant (S-D) lo-gic. The best known model of this approach isthe Gallouj-Weinstein competence-based model(Gallouj and Weinstein 1997) that represents aproduct or a service as a system of (provider)competences (PCi), technical characteristics (PTi),and final characteristics (Oi), where the serviceoutcome (Oi) is resulted from the interactionsbetween the customer competences (CCi) andthe provider’s competences (PCi) and technicalcharacteristics (PTi). Service innovations thusconsist of changes in one or more of these ele-ments. Provider competences PCi are then thedirect mobilisation of service personnel compet-ences (i.e., without any technological mediation).PTi are knowledge, competences embodied intangible (such as front- and back-office character-istics) or intangible (i.e., codified and formalisedcompetences such as job analysis methods. Afundamental characteristic of service activitiesis client participation (in various forms) in theproduction of the service (Gallouj 2002).

5.3 Service Innovation Process andManagement

Service innovation competence is a crucial oper-ant resource for the firm’s competitive advantage.Service innovation practice depends critically ona streamlined and flexible process for internaland external resource coordination and integra-tion to achieve effective and efficient customervalue co-creation. Service innovation process,also known as new service development, gener-ally (Engel et al. 2006; Thomke 2003) consists offive phases:

• Create ideas – this phase defines the idea, itsscope and business benefits

• Evaluate and select ideas – this phase prior-itises the portfolio of ideas and develops theselected idea into a (low cost low risk) experi-ment to test its feasibility; go/no go decision ismade quickly to speed up the chance of identi-fying a feasible idea (or conversely the rate offailures of infeasible ideas)

• Plan, design, develop and implement ideas –this phase takes the feasible idea through arigorous service development lifecycle

• Commercialise the ideas – this phase launchesthe service

• Review the impacts – this phase reviews theresults of the innovation to improve currentperformance and as a feedback for future pro-cess improvement

However, as alluded to in the design practicesframework (Sect. 3.1), in the digital world this in-novation process would not necessarily occurin a purely linear (predictive) manner, ratherit would tend to be circularly iterative, akin to"agile (emergent) development".

Research on service innovations has highlightedthe critical importance of the front-end stages ofnew service development: idea generation, ideascreening and concept development – collect-ively known as the fuzzy front-end (Alam 2006).Customer involvements in the front-end stagesof a service innovation process are important


82 Eng K. Chew

so as to reduce the fuzziness (Alam 2006). Ser-vice innovation may be incremental for steadybusiness growth – through exploitation of exist-ing competences (O’Reilly and Tushman 2008);or radical for new growth idea (Anthony et al.2008), which could become a new growth plat-form (Laurie et al. 2006) – through exploration ofnew competences/capabilities (O’Reilly and Tush-man 2008). But the exploratory activities must bebuffered from exploitative activities to ensure co-existence (Benner and Tushman 2003), creatinga so-called ambidextrous organisation capableof both exploitative and exploratory innovationssimultaneously.

Companies are also increasingly leveraging in-novative ideas from outside the firms using anopen innovation process (Chesbrough 2003). Thismeans the firm needs to engage customers, part-ners, suppliers, regulators, and even competitorsto co-generate creative ideas, co-produce serviceofferings and co-create value in a continual non-linear process of service innovation, which sup-ports direct interactions with the customers tomatch innovations with customers needs (Ches-brough 2011). The aim of customer participation,as described in the next section, is to co-create a"unique personalised customer experience" (Pra-halad and Krishnan 2008).

6 Customer ParticipationCentral to discovering service innovation op-portunities is "knowing how customers definevalue" (Bettencourt 2010). As service value is al-ways determined by the customer, new creativeideas must be developed from the customer’soutside-in view (Edvardsson et al. 2007; Payneet al. 2008). Indeed, successful firms are co-optingcustomer involvement in service and value co-creation (Prahalad and Ramaswamy 2000). Cus-tomer participation is equally essential to boththe ’old’ physical and ’new’ digital service worlds.However, involving customer in co-productionof a service process is often confronted with con-flicting design requirements. For example, scale-economy or efficiency requirements would de-mand service standardisation, while personalised

service experience requirements would demandservice variability tailored to individual prefer-ences. In general, customer participation is in-herently a source of variability since each cus-tomer has different capabilities and must learnhow to interact with the service process (Mettersand Marucheck 2007). The concept of customerefficiency is therefore a critical requirement ofservice process design to denote the customer’sability to participate in self service or coproduceservice (Metters and Marucheck 2007; Xue andHarker 2002) – for instance the user innovationtoolkit (Hippel 2001). Similarly, customer vari-ability is, thus, a design variable which can bemanaged to improve both service quality andefficiency (Metters and Marucheck 2007).

Firms compete through service by collaborat-ing (i.e., co-produce offering and co-create value)with customers and network partners to enhanceknowledge (Lusch et al. 2007). This requires thefirm to possess absorptive capacity (Zahra andGeorge 2002) in order to absorb new informa-tion and knowledge from customers and partnersto comprehend from the external environmentsthe important trends and know-how which, inturn, give them the ability to adapt/adjust to thecomplex, dynamic, and turbulent external envir-onments. Firms that draw on the knowledge oftheir customer base can capitalise on customercompetencies for use during the course of theirinnovation activities (Blazevic and Lievens 2008).

Customer participation or involvement in serviceinnovation can take place at various phases ofthe new service development process (Alam 2006;Chesbrough 2011). Customer participation or in-tegration can be conceptualised as the incorpora-tion of resources from customers into the servicedevelopment processes of a company (Moeller2008). This would include participating in pro-ducing and delivering the service (Dong et al.2008). Business has to develop an adaptive or-ganisational model where customer involvementand innovation is persistent and inherent in theentire service lifecycle – such that the distinc-tion between customers and employees becomes


blurred (Oxton 2008). This organisational modeloperates as a network of relationships based onthe principles of alignment, transparency, iden-tity (reputation) (Oxton 2008).

Customer participation towards creating person-alised experience (Prahalad and Krishnan 2008)typically follows a five-stage iterative approach:1) establishment of antecedent conditions forcustomer to participate; 2) development of mo-tivations or customer benefits; 3) cost-benefitevaluation; 4) activation of co-creation processby choosing the stages of the "production-con-sumption" activity chain; and 5) evaluation ofthe effectiveness of the co-creation strategiesagainst the cost-benefit analysis (Etgar 2008). Itis prudent for the provider to institute a continu-ous learning process with the customer from theco-creation experience to improve their service-usage competence. Learning enhances the cus-tomer’s competence in seamlessly integratingthe value proposition with their lives, objectivesand aspiration (Payne et al. 2008). Organisationallearning about customer’s value creation pro-cesses deepens customer insights. Organisationallearning is a crucial process for nurturing the pro-vider’s collaborative competence to improve theprovider’s innovation capability and competitiveadvantage (Edmondson 2008).

The increased digitalisation of services in theinternet era is creating new opportunities forknowledge coproduction between customers andthe provider (Blazevic and Lievens 2008). In adigital world, customers may take on three dif-ferent roles for knowledge coproduction-passiveuser, active informer, and bidirectional creator-each with distinctive declarative and proceduralcharacteristics, and distinct impacts on the threeinnovation tasks of detection, development, anddeployment (Blazevic and Lievens 2008). The di-gital world also facilitates customer participationin recovery from service failure. This may varyin degrees from firm recovery, joint recovery, tocustomer recovery (Dong et al. 2008). This wouldrequire higher levels of role clarity, but it alsotends to enhance satisfaction with the service

experience, perceived value in future co-creation,and intention to co-create in the future (Donget al. 2008).

7 Community-based Innovation

The advent of social media and clouds-based ser-vices has led many firms globally, as part of im-plementing their social strategies, to directly en-gage with their customers online across a broadrange of activities (such as marketing, customercare, etc.) to co-create value for mutual bene-fits. This has evolved from a relatively straight-forward traditional online customer service plat-form to a more sophisticated community basedinnovation (CBI) which requires a new set oforganisational capabilities that interact and in-tegrate with those of the customers themselves(Fuller et al. 2006).

CBI is defined as a new online service innovationprocess that fully engages the firm’s customercommunity from ideation phase right throughto the test and launch phase of New Service De-velopment. The community members becomethe sources of new service ideas as well as theco-creators and evaluators of the service designs.The most common CBI user/customer archetypeis called the "lead users" – who are highly know-ledgeable of the firm’s products/service and have’job’ (problem) needs that are ahead of all otheruser groups in a given market. Lead users areallowed to design (using interactive toolkits pro-vided by the service provider) their own products/service by trial-and-error according to their wantsand needs. Their creativity and problem-solvingskills (competencies) using the toolkits (providercompetencies) will produce the ’ideal’ solutionsto match their problems (the ’jobs’ to be done) –for instance, Peugeot’s "Retrofuturism" cardesigns were produced using CBI1 (Fuller et al.2006). Two other user archetypes are also com-mon: the "insiders" who are strongly associatedin the community and highly involved in thetopic; the "devotee" who are highly involved with

1www.peugeot-avenue.com.

www.peugeot-avenue.com


84 Eng K. Chew

the topic but not very much related with the com-munity. CBI communities could be selected onthe basis of the exchanged content, professional-ism, traffic volume, and number of participantsinteracting with each other (Fuller et al. 2006).Users could be accessed directly or more oftenthey recommend access via a trustworthy mem-ber of the community or via the webmaster toincrease acceptance. Feedback to users on theirinput is regarded critical as is getting users’ feed-back on their participation experience and theirwillingness and expectations to participate againin future virtual product/service developmentprojects (Fuller et al. 2006).

Community members engagement in CBI canbe fostered and sustained in a three-step pro-cess: (1) understand consumer needs and mo-tivations; (2) promote community participation,including encourage content creation, cultivateconnections, and create enjoyable experiences;and (3) motivate cooperation, including mobil-ising member-leaders, inspiring idea creation andselection via a panel/polling (Porter et al. 2011).Community engagement is motivated intrinsic-ally by the value created when community spon-sors help user-members meet their needs withtheir virtual community. So the community spon-sor’s judicious and targeted efforts to encouragemembers to act in ways that create greater valuefor themselves and for the firm are crucial tosuccess (Porter et al. 2011). Members’ "embed-dedness" (willingness to act in value-creatingways toward a community sponsor) and "em-powerment" are seen to be fundamental to driv-ing cooperative, engaging behaviour from thecommunity members (Porter et al. 2011). This, inturn, would require the community sponsor tounderstand the needs of its community members,build trust with and create value for its members(Porter et al. 2011). CBI tends to focus on firm-community (one-to-many and many-to-one) col-laboration. More recently, new social strategiesare being proposed that seek to reduce companycosts and/or increase customer willingness to payby helping the community to meet online and

strengthening their relationships – that is focuson many-to-many social activities between com-munity members as exemplified by eBay’s GroupGift (Piskorski 2011).

8 Strategic Management for InnovationSuccess

Innovative service firms have strong commit-ment to innovation from top management backedby well structured innovation processes and gov-ernance together with the aligned culture andsystems, and the attendant prioritised resourcesallocated to innovation efforts. In service in-novation "it is not the service itself that is pro-duced but the pre-requisites for the service" (Ed-vardsson and Olsson 1996). Due to services’ real-time production, new service development wouldrequire modifications of the service delivery pro-cess and changes in frontline employees’ skills.This would require strong fit between the newservice and existing systems; and close alignmentbetween the customer-service-focused front-endand the operational-excellence-focused back-endsystems.

But despite its strategic importance, service in-novation is notoriously difficult to accomplish(Dorner et al. 2011). This could be attributed tosuch managerial deficiencies as: lack of ability toprotect services hinders investment; lack of clear"organisational anchoring" of service innovationactivities; lack of systematic innovation process;lack of customer participation; and "bad ideas notconsistently eliminated" (Chandy and Tellis 1998).So managers need to be vigilant in all innovationstages to assess ideas against the company’s stra-tegic goals and market needs in order to determ-ine their commercial viability. Further, managersneed to focus on people (evolving competencesin line with changing customer value expecta-tions) and structural support (systematic newservice development process supported by spe-cific innovation tools, multi-disciplinary teams,the availability of resources, market testing andmarket research) to ensure successful service in-novation (Dorner et al. 2011).


Service innovation is technology-enabled butmore human-centred and process-oriented.Therefore, the "envisioning, energising and en-abling" capabilities, sound communication/co-ordination, and reducing intra-organisational con-flicts and power struggle have been identifiedas fundamental and very critical for new ser-vice development to minimise organisational in-ertia/resistance (Nijssen et al. 2006). Innovativefirms commonly possess "willingness to canni-balise" mindset and capability – i.e., willingnessto make obsolete its existing products/services,prior investments, and/or existing organisationalcapabilities (Chandy and Tellis 1998; Nijssen et al.2006). These innovative organisations are said topossess ambidexterity capable of pursuing sim-ultaneous exploitative and exploratory innova-tions. An ambidextrous organisation "requires acoherent alignment of competencies, structuresand cultures to engage in exploration, a contrast-ing congruent alignment focused on exploitation,and a senior leadership team with the cognit-ive and behavioural flexibility to establish andnurture both" (O’Reilly and Tushman 2008).

9 Conclusion

Service innovation is focused on creating cus-tomer value, and service is about relationshipwith the customer. Customer co-creates valuewith the provider by integrating his/her com-petences/capabilities with those of the provider.Thus customer productivity is as important asthat of the provider in service provision as itimpacts directly the service experience. Increas-ingly, in a digital world, customer and member-community participation across the firm’s en-tire service innovation lifecycle is becoming acritical innovation strategy for sustained valueco-creation. It has become a core and distinct-ive organisational capability for service organ-isations to develop and adapt in line with theevolving external environments and the custom-ers’ increasingly mature service competences.

Service innovation is technology-enabled butmore human-centred and process-oriented. This

is accentuated by the design practices frame-work for service innovation which serves as afoundation for systematic service conceptualisa-tion, design, architecture and innovation. Serviceinnovation commercialisation is contingent onmindful alignment of the firm’s service strategy,service concept and business model. Firm needscollaborative, absorptive capacity and dynamiccapabilities (including organisational learningprocesses) to continuously adapt its service in-novations with the changing external environ-ments including the value networks to which it isconnected. From strategic management perspect-ive, the firm needs to be ambidextrous capableof pursuing exploitative and exploratory serviceinnovations simultaneously to create sustainedvalue for itself and its customers.

Acknowledgements

This paper is based on and provides an extensionto the author’s earlier work described in parts ofChew and Gottschalk (2013).

References

Alam I. (2006) Removing the fuzziness from thefuzzy-end of service innovations through cus-tomer interactions. In: Industrial MarketingManagement 35(4)

Anthony S. D., Johnson M. W., Sinfield J. V.(2008) Institutionalizing innovations. In: MITSloan Management Review 49(2)

Arnould E. J. (2008) Service-dominant logic andresource theory. In: Journal of the Academyof Marketing Science 36, pp. 21–24

Benner M. J., Tushman M. L. (2003) Exploitation,exploration and process management: theproductivity dilemma revisited. In: Academyof Management Review 28

Bettencourt L. A. (2010) Service innovation: howto go from customer needs to breakthroughservices. McGraw-Hill, New York

Bitner M. J., Ostrom A. J., Morgan F. N. (2008)Service blueprinting: a practical techniquefor service innovation. In: California Man-agement Review 50(3), pp. 66–94


86 Eng K. Chew

Blazevic V., Lievens A. (2008) Managing innov-ation through customer co-produced know-ledge in electronic services: an exploratorystudy. In: Journal of the Academy of Market-ing Science 36, pp. 138–151

Bolton R. N., Grewal D., Levy M. (2007) Sixstrategies for competing through service: anagenda for future research. In: Journal of Re-tailing 83(1), pp. 1–4

Casadesus-Masanell R., Ricart J. E. (2010) Fromstrategy to business models and onto tactics.In: Long Range Planning 43, pp. 195–215

Chandy R. K., Tellis G. J. (1998) Organizing forradical product innovation: the overlookedrole of willingness to cannibalize. In: Journalof Marketing Research 35(4), pp. 474–487

Chesbrough H. (2003) A better way to innovate.In: Harvard Business Review 81(7), pp. 12–13

Chesbrough H. (2011) Bringing open innovationto services. In: MIT Sloan Management Re-view 52(2), pp. 85–90

Chesbrough H., Davies A. (2010) Advancingservice innovations. In: Maglio P. P., Kiel-iszewski C. A., Spohrer J. C. (eds.) Hand-book of Service Science. Springer, New York,pp. 579–601

Chew E., Gottschalk P. (2013) Knowledge drivenservice innovation and management: ITstrategies for business alignment and valuecreation. IGI Global, Hershey, PA, USA.

Clark G., Johnston R., Shulver M. (2000) Exploit-ing the service concept for service design anddevelopment. In: Fitzsimmons J. A., Fitzsim-mons M. J. (eds.) New Service Design. Sage,Thousand Oaks, CA, pp. 71–91

Dong B., Evans K. R., Zou S. (2008) The effectsof customer participation in co-created ser-vice recovery. In: Journal of the Academy ofMarketing Science 36

Dorner N., O. G., Gebauer H. (2011) Service in-novation: why is it so difficult to accomplish?In: Journal of Business Strategy 32(3), pp. 37–46

Edmondson A. C. (2008) The competitive im-perative of learning. In: Harvard BusinessReview 86(7/8)

Edvardsson B., Olsson J. (1996) Key concepts fornew service development. In: Service Indus-tries Journal 16(2), pp. 140–164

Edvardsson B., Gustafsson A., Roos I. (2005) Ser-vice portraits in service research: a criticalreview. In: International Journal of ServiceIndustry Management 16(1), pp. 107–121

Edvardsson B., Gustafsson A., Enquist B. (2007)Success factors in new service developmentand value creation through services. In: SpathD., Fahnrich K.-P. (eds.) Advances in Ser-vices Innovations. Springer, Berlin Heidel-berg, pp. 166–183

Eisingerich A. B., Rubera G., Seifert M. (2009)Managing service innovation and interorgan-izational relationships for firm performance:to commit or diversify? In: Journal of ServiceResearch 11(4), pp. 344–356

Engel J. F., Thompson A. M., Nunes P. F., LinderJ. C. (2006) Innovation unbound. In: Accen-ture Publication 1, pp. 28–37

Etgar M. (2008) A descriptive model of the con-sumer co-production process. In: Journal ofthe Academy of Marketing Science 36

Fitzsimmons J. A., Fitzsimmons M. J. (2010) Ser-vice Management: Operations, Strategy, In-formation Technology., 7th ed. McGraw-Hill,Irwin, New York, NY

Fixson S. K. (2005) Product architecture assess-ment: a tool to link product, process, and sup-ply chain design decisions. In: Journal of Op-erations Management 23(3/4), pp. 345–369

Froehle C. M., Roth A. (2004) New measure-ment scales for evaluating perceptions of thetechnology-mediated customer service exper-ience. In: Journal of Operations Management22(1), pp. 1–22

Fuller J., Bartl M., H. E., Muhlbacher H. (2006)Community based innovation: how to integ-rate members of virtual communities intonew product development. In: ElectronicCommerce Research 6, pp. 57–73

Fynes B., Lally A. M. (2008) Innovation in ser-vices: from service concepts to service experi-ences. In: Hefley B., Murphy W. (eds.) ServiceScience, Management and Engineering Edu-


cation for the 21st Century. Springer, NewYork, pp. 229–333

Gadrey J., Gallouj F. (2002) Productivity, Innov-ation and Knowledge in Services: New Eco-nomic and Socio-economic Approaches. Ed-ward Elgar, Cheltenham, UK

Gallouj F. (2002) Innovation in the service eco-nomy: the new wealth of nations. EdwardElgar, Cheltenham, UK

Gallouj F., Weinstein O. (1997) Innovation inservices. In: Research Policy 26, pp. 537–556

Glushko R. J., Tabas L. (2009) Designing ser-vice systems by bridging the ‘front stage’and ‘back stage’. In: Information Systems, E-business Management 7, pp. 407–427

Goldstein S. M., Johnston R., Duffy J., Rao J.(2002) The service concept: the missing linkin service design research? In: Journal of Op-erations Management 20(2), pp. 121–134

Grawe S. J., Chen H., Daugherty P. J. (2009) Therelationship between strategic orientation,service innovation, and performance. In: In-ternational Journal of Physical Distribution& Logistics Management 39(4), pp. 282–300

Hastings H., Saperstein J. (2013) A practice-driven service framework for value creation.In: 2013 IEEE International Conference onBusiness Informatics. Piscataway, NJ, USA

Helkkula A., Kelleher C., Pihlstrom M. (2012)Characterizing value as an experience: im-plications for service researchers and man-agers. In: Journal of Service Research 15(1),pp. 59–75

Hertog P. d. (2002) Co-producers of innovation:on the role of knowledge-intensive businessservices in innovation. In: Gadrey J., GalloujF. (eds.) Productivity, Innovation and Know-ledge in Services: New Economics and Socio-Economic Approaches. Edward Elgar, Chel-tenham, UK, pp. 223–255

Heskett J. L., Jones T. O., Loveman G. W., Sas-ser W. E., Schlesinger L. A. (2008) Puttingthe service-profit chain to work. In: HarvardBusiness Review 86(7/8)

Hippel E. v. (2001) Perspective: user toolkits forinnovation. In: The Journal of Product Innov-

ation Management 18, pp. 247–257Holmlid S., Evenson S. (2008) Bringing service

design to service sciences, management andengineering. In: Hefley B., Murphy W. (eds.)Service Science, Management and Engineer-ing Education for the 21st Century. Springer,New York, pp. 341–345

Johannessen J. A., Olsen B. (2010) The futureof value creation and innovations: aspectsof a theory of value creation and innovationin a global knowledge economy. In: Interna-tional Journal of Information Management30, pp. 502–511

Laurie D. L., Doz Y. L., Sheer C. P. (2006) Creat-ing new growth platforms. In: Harvard Busi-ness Review 84(5)

Lovelock C., Gummesson E. (2004) Whitherservices marketing?: in search of a newparadigm and fresh perspectives. In: Journalof Services Research 7, pp. 20–41

Lusch R. F., Vargo S. L., O’Brien M. (2007) Com-peting through service: insights from servicedominant logic. In: Journal of Retailing 83(1),pp. 5–18

Lusch R. F., Vargo S. L., Tanniru M. (2009) Ser-vice, value networks and learning. In: Journalof the Academy of Marketing Science 38(1),pp. 19–31

Madhavaram S., Hunt S. D. (2008) The service-dominant logic and a hierarchy of oper-ant resources: developing masterful oper-ant resources and implications for market-ing strategy. In: Journal of the Academy ofMarketing Science 36, pp. 67–82

Maglio P. P., Spohrer J. (2008) Fundamental ofservice science. In: Journal of the Academyof Marketing Science 36, pp. 18–20

Maglio P. P., Srinivasan S., Kreulen J. T., SpohrerJ. (2006) Service systems, service scientists,SSME and innovation. In: Communicationsof the ACM 49(7), pp. 81–85

Maglio P. P., Vargo S. L., Caswell N., Spohrer J.(2009) The service system is the basic abstrac-tion of service science. In: Information Sys-tems E-Business Management 7(4), pp. 395–406


88 Eng K. Chew

Metters R., Marucheck A. (2007) Service manage-ment – academic issues and scholarly reflec-tions from operations management research-ers. In: Decision Sciences 38(2), pp. 195–214

Moeller S. (2008) Customer integration – a keyto an implementation perspective of serviceprovision. In: Journal of Service Research11(2), pp. 197–210

Nijssen E. J., Hillebrand B., Vermeulen P. A.M., Kemp R. G. M. (2006) Exploring productand service innovation similarities and differ-ences. In: International Journal of Researchin Marketing 23

Normann R., Ramirez R. (1993) From value chainto value constellation: designing interactivestrategy. In: Harvard Business Review 71,pp. 65–77

Ordanini A., Parasuraman A. (2011) Service in-novation viewed through a service domin-ant logic lens: a conceptual framework andempirical analysis. In: Journal of Service Re-search 14(1), pp. 3–23

O’Reilly C. A., Tushman M. L. (2008) Ambidex-terity as a dynamic capability: resolving theinnovator’s dilemma. In: Research in Organ-izational Behavior 28

Osterwalder A., Pigneur Y. (2005) Clarifyingbusiness models: origins, present, and futureof the concept. In: Communications of theAssociation for Information System 16, pp. 1–25

Oxton G. (2008) An integrated approach to ser-vice innovation. In: Hefley B., Murphy W.(eds.) Service Science, Management and En-gineering Education for the 21st Century.Springer, New York, pp. 97–105

Patricio L., Fisk R. P., Cunba J. F. (2008) Design-ing multi-interface service experiences: theservice experience blueprint. In: Journal ofService Research 10(5)

Payne A. F., Storbacka K., Frow P. (2008) Man-aging the co-creation of value. In: Journal ofthe Academy of Marketing Science 36, pp. 83–96

Piskorski M. J. (2011) Social strategies that work.In: Harvard Business Review 89(11), pp. 117–

122Porter C. E., Donthu N., MacElroy W. H., Wydra

D. (2011) How to foster and sustain engage-ment in virtual communities. In: CaliforniaManagement Review 53(4), pp. 80–110

Prahalad C. K., Krishnan M. S. (2008) New age ofinnovation: driving cocreated value throughglobal networks. McGraw-Hill, New York

Prahalad C. K., Ramaswamy V. (2000) Co-optingcustomer competence. In: Harvard BusinessReview 78(1), pp. 79–87

Roth A. V., Menor L. J. (2003) Insights intoservice operations management: a researchagenda. In: Production and Operations Man-agement 12(2), pp. 145–164

Schneider B., Bowen D. E. (2010) Winning theservice game. In: Maglio P. P., Kieliszewski,C. S. J. (eds.) Handbook of Service Science.Springer, New York, pp. 31–59

Spohrer J., Maglio P. P., Bailey J., Gruhl D. (2007)Steps toward a science of service systems. In:IEEE Computer 1, pp. 70–77

Teece D. J. (2007) Explicating dynamic capab-ilities: the nature and microfoundations of(sustainable) enterprise performance. In: Stra-tegic Management Journal 28, pp. 1319–1350

Teece D. J. (2010) Business models, businessstrategy and innovation. In: Long Range Plan-ning 43, pp. 172–194

Thomke S. (2003) R&D comes to services – Bankof America’s pathbreaking experiments. In:Harvard Business Review 81(4), pp. 71–79

University of Cambridge, IBM (2007) Succeed-ing through Service Innovation: A DiscussionPaper. University of Cambridge Institute forManufacturing, Cambridge, United Kingdom

Vargo S. L., Lusch R. F. (2004) Evolving to a newdominant logic for marketing. In: Journal ofMarketing 69, pp. 1–17

Vargo S. L., Lusch R. F. (2008) Service-dominantlogic: continuing the evolution. In: Journal ofthe Academy of Marketing Science 36, pp. 1–10

Vargo S. L., Maglio P. P., Akaka M. A. (2008)On value and value co-creation: a servicesystems and service logic perspective. In:


European Management Journal 26, pp. 145–152

Voss C. A., Hsuan J. (2009) Service architectureand modularity. In: Decision Sciences 40(3),pp. 541–569

Womack J. P., Jones D. T. (2005) Lean con-sumption. In: Harvard Business Review 83(3),pp. 58–68

Xue M., Harker P. T. (2002) Customer efficiency:Concept and its impact on e-business man-agement. In: Journal of Service Research 4(2)

Zahra S. A., George G. (2002) Absorptive capa-city: a review, reconceptualization, and ex-tension. In: Academy of Management Review27(2), pp. 185–203

Zomerdijk L. G., Voss C. A. (2010) Service designfor experience-centric services. In: Journal ofService Research 13(1), pp. 67–82

Zott C., Amit R. (2010) Business model design:an activity system perspective. In: LongRange Planning 43, pp. 216–226

Eng K. Chew

School of Systems, Management and LeadershipUniversity of [email protected]


90 Stéphane Marchand-Maillet and Birgit Hofreiter

Stéphane Marchand-Maillet and Birgit Hofreiter

Big Data Management and Analysis for BusinessInformatics

A Survey

Modern communication networks have fueled the creation of massive volumes of data that may be valued asrelevant information for business activities. In this paper, we review technologies for enabling and empoweringbusiness activities, leveraging the content of this big data. We distinguish between data- and user-relatedtechnologies, and study the parallel brought by the overlap of these categories. We show how the trend ofBig Data is related to data security and user privacy. We then investigate automated ways of performingdata analysis for Business Intelligence. We finally review how groups of users may be seen as a workforce inbusiness through the notion of human computation or crowdsourcing, associated with the notions of trust andreputation. We conclude by discussing emerging trends in the domain.

1 Introduction

Progress in Business Informatics aim to developbusiness administration using computational andinformation technologies. As such, business in-formatics may use any method providing a tech-nology useful for its end purpose.

Modern business activities essentially rely onan accurate management of knowledge (oftenreferred to as Business Intelligence). The devel-opment of communication technologies and thewide-spread and ubiquity of communication net-works have created an opportunity for gather-ing and analysing data in view of deriving use-ful knowledge. Hence, business informatics isprimarily supported by data management anddata analysis technologies. In addition, users anduser groups remain at the center of any busi-ness. They may assist performing data analysisas much as benefiting from it.

In this paper, we review and analyse the mainenabling technologies in business informatics.We explore as thoroughly as possible the inform-ation landscape in which business informaticsoperates, to understand the aspects and their

characteristics, potential risks and benefits. User-generated data is considered a potentially richsource of information for business and user be-haviors are modeled using this data. Users anddata are therefore two inter-related main actorswithin this landscape that we explore via thesetwo perspectives, as illustrated in Fig. 1.

Data Users

Storage and access

Big Data

Visual

Analytics

Social Networks

(Web 2.0)

Management

Analysis

Machine Intelligence Human Computation

Knowedge

Management

Ontologies

Security Privacy

Trust

Communities

Social Knowledge

Figure 1: A classification of domains for enablingand empowering technologies in business informatics.Square boxes indicate technical domains, whereas cir-cular items relate to multi-disciplinary binding fieldsof study

Investigating data-related aspects allows to un-derstand the technical infrastructure that should

Enterprise Modelling and Information Systems ArchitecturesVol. 9, No. 1, June 2014Big Data Management and Analysis for Business Informatics 91

be set up and sustained, from base data collec-tion and housing to sophisticated data analysis.In parallel, studying user-related issues allows tomodel the user and his community, as originatorof this data. We therefore distinguish betweendata- and user-related technologies, although thesplit is somewhat artificial since these techno-logies generally overlap. We further look intopassive management technologies (whose mainaim is not to create any knowledge) against act-ive analysis technologies (that transform datainto knowledge). The extend of our review issymbolised in Fig. 1. Section 2 reviews the hous-ing and preservation of data, in relation to thecurrent challenge of Big Data. In Sect. 3, we re-view automated technologies for data analysis.These are crucial for their aspect of scalability,since any necessity for user intervention wouldcreate prohibitive costs at large scale. As a mir-ror to the data-related sections, Sect. 4 reviewsuser-related strategies, from preserving user pri-vacy to exploiting the force and intelligence ofthe crowd. We discuss the future potential of re-viewed technologies and foreseeable extensionsin Sect. 5.

2 Data management

While information and communication networkshave triggered the creation of an overwhelmingmass of data, they have also created opportunit-ies to monitor and mine this data, thus augment-ing drastically the volume of contextual informa-tion potentially available. Massive collections ofdigital documents are made available, either pub-licly on the Web or in private networks relatedto companies, workgroups or social associations.Data may be very diverse and arise from anyform of information exchange. Examples of thisinclude:

• Textual: Web pages (personal, professional,from individuals, groups or companies), emails,blogs posts, news feeds, exchanges over socialnetworks;

• Multimedia: photos, videos, music, audiovisualblogs, ...;

• Process data: sensing data (GPS, weather, trafficmonitoring, ...), computation (sociological, sci-entific, financial trends, ...);

• Logs: traces of human interaction with sys-tems (information, e-commerce, entertainment,...), logs of machine-machine communications(web services, distributed computing, ...).

Before considering data analysis, a choice is tobe made on the form of data housing, if any.The emergent paradigm of big data (Sect. 2.1) isaddressing some choices there. In turn, data pre-servation and access immediately open securityissues, reviewed in Sect. 2.2.

2.1 Big data

Every study on the topic shows clearly that thevolume of data created by individuals and com-panies is growing exponentially (see, e.g., Ma-nyika et al. 2011). In parallel, analysts predictsuccess to anyone who will exploit this data ac-curately, thus implicitly supporting the mean-ingfulness of this data. However, this data iseverything unlike what companies are used todeal with. It is unstructured and redundant andpotentially noisy or corrupted. Every piece ofthe data may be seen as noise that would pollutea database of clean and structured data. There isnevertheless a clear intuition that the mass com-pensates for the defects of the pieces. A globalpicture of the data should contain informationthat could be exploited to many ends. This isthe challenge of the recent trend identified as bigdata (Big Data: Science in the Petabyte Era 2008;ERCIM News Special Theme: Big Data 2012).

Big data has initially been characterised by its“three Vs” (Fayyad 2012), mostly addressing itstechnical specificity:

• Volume: In itself, the volume of this data is anissue. It surpasses many of the simple storagestrategies classically used. At this scale andevolution rate, it is hardly possible to structureand clean the data, for both technical and costreasons;



• Variety: In order to transform data into know-ledge, its multiple facets should be taken intoaccount (see Sect. 3). The data in questiontherefore encompasses a high diversity in itscontent, format, structure and interpretation.Again, this goes against most principles of clas-sical data management and storage paradigmswhere the structure of the data should be un-derstood and stable;

• Velocity: One of the main characteristics of bigdata is the pace at which it is generated andat which it evolves and gets obsolete. In otherwords, this data inherently bears a strong tem-poral dimension. Usage logs, trends, news,are all content that have a strong interest intheir immediate history and quickly declineinto useless or even polluting data. However,this data may also have a behavioral interestat long-term on a more global temporal scale(e.g., Morrison et al. 2012).

These technical factors prevent the use of the typ-ical database models (e.g., a relational structuremade usable via SQL) and impose to move ontomore flexible, agile and scalable structures (in-cluding the NoSQL trend1 advocating for schema-free storage or the MapReduce model (Dean andGhemawat 2004; Mohamed andMarchand-Maillet2012)2 to support indexing, e.g., via the agnosticname-value pair model). Decentralised storageand processing systems (a.k.a Peer-to-Peer sys-tems or the Cloud) rely on structures havingthese characteristics to make the data safe, ubi-quitous, and accessible (“Anything, Anywhere,Anytime”).

In practice, the current appeal for big data hasmapped into new functions coined as data scient-ists, i.e, data analysts able not only to performtechnical operations on the data such as prepara-tion, cleaning, compaction, etc, but also to contex-tualise the data and read it with all surroundingparameters (e.g., social context, as detailed inSect. 4.3).

1Web-scale databases: http://nosql-database.org2The Apache Hadoop library: http://hadoop.apache.org/

Many “Vs” have been added to big data (e.g., Vi-ability, Veracity (Vossen 2013), Volatility, ...) butthe main “V” business is concerned with is

• Value: The question is “how to make value outof this large, complex and unstable stream ofdata?”

There are many answers to that question, includ-ing:

• By better understanding actions and behaviorof its customers, traced by the log of their ac-tions, a company will be able to offer betterand more relevant services;

• By better understanding the context withinwhich it operates, characterised by the min-ing of environmental factors, a company willlower its risks.

The first common step is always to transformdata into knowledge, partly thanks to the techno-logies reviewed in the next sections. One findsreports on success stories of big data analytics3

relating how such insurance company could fine-tune its risk model using deep data analysis, in-cluding exploiting inferred customer social re-lationships (which in turn poses questions onprivacy - see Sect. 4.1. Looking at business asa permanent complex constrained optimisationproblem, where the right balance should be found(“price vs volume, cost of inventory vs the chanceof a stock-outrisk”3, etc), the success of big data isin providing insights into how to rationalise de-cisions on these tradeoffs. In that case, the noisein the data refers to any potential inconsistencyin data patterns, unintentional (e.g., failures inprocess or communication), or intentional (e.g.,spam (Mukherjee et al. 2012)).

It should be emphasised there that as much asthere may be some benefit for a company to ex-tract knowledge from the data, there is an inher-ent risk that this data is used by an adversarialparty in many ways. These includes industrial

3e.g., http://www.forbes.com/sites/mckinsey/2012/12/03/big-data-advanced-analytics-success-stories-from-the-front-lines

http://nosql-database.org

http://hadoop.apache.org/

http://www.forbes.com/sites/mckinsey/2012/12/03/big-data-advanced-analytics-success-stories-from-the-front-lines




espionage for adversarial reasons and unfair com-petition, signals intelligence collection and ana-lysis4 for (state) security reasons and customerprivacy breach or exposure5, either intended oraccidental. This then forces to ensuring the secur-ity of the data and the privacy of the user, whichwe study next.

2.2 Data security

Securing data first aims at preserving its integ-rity and confidentiality, while not imposing con-straints on its availability to authorised parties.Issues linked to its authenticity and accountabilityare related to its integrity, while data access ischaracterised by non-repudiation and reliability.

Data security is related to data usage and there,related services include user authentication, userauthorisation, access accountability and user reli-ability. This finally mirrors to user privacy andtrust, studied in detail in Sect. 4.1 and Sect. 4.2,respectively. The relationship between data se-curity and user privacy is established via “guar-anteeing privacy by securing access to privatedata”. On a sociological level, the Security Cul-ture (Alnatheer et al. 2012) is defined as the levelof belief and expectations members of a group(e.g., an organisation or company) have regard-ing security. This is valid at all levels of theorganisation and, in Ruighaver et al. (2007), itis demonstrated that the effectiveness of opera-tional security policies in an organisation is posit-ively correlated with the belief that the top man-agement (decision makers) have in these policies.Greene and D’Arcy (2010) verify empirically thehypothesis that security culture (i.e., beliefs) andjob satisfaction lead to a increased security beha-vior in the organisation.

A further distinction regarding the integrationof data security is to be made between the pub-lic and private sector. The absence of economicmarkets for final product outputs in the public

4e.g., the Echelon Network (2001) over communicationnetworks, or the Prism Initiative (2013) over the Web

5e.g., the case of “AOL user 927” (2006)

sector and the associated reliance on governmentfinancial resources place public agencies understrong political influence. As a result, informa-tion management systems in public organisationsemphasise more the environmental factors ratherthan internal characteristics from the organisa-tion. As demonstrated in Conklin (2007); Wang(2009), these differences play an important role inthe diffusion of technology in e-government set-tings. In particular, decisions made on informa-tion security management in public organisationsdo not always follow technological rationales(Ruighaver et al. 2007).

Technologically, the challenge is to define secur-ity strategies in the Information Management Sys-tem that will support business processes (see, e.g.,Diesburg and Wang 2010 for technical surveyson digital data security). As given in Place andHyslop (1982), Information management focuseson “plans and activities that need to be performedto control an organisation’s records”. Here, secur-ity should “ensure the continuity and minimisebusiness damage by preventing and minimisingthe impact of security incidents” (Solms 2006,2010). Authors of the latter references structurethe evolution of security policies into severalwaves (illustrated in Fig. 2) showing the focus ofevery development phase, from purely technolo-gical to management-related concerns.

Technical Wave

Governance Wave

Management Wave

Cyber Security Wave

Institutional Wave

- 1983 1985 - 1990 1995 - 2000 2000 - 2005 2006 -

Figure 2: The 5 waves of Information Security (createdafter Solms 2010). The drift from technical to societalissues is clearly visible

From its origins (first wave), information securityhas been seen from a technical perspective. With



the stability of the technical solutions, the ques-tion has moved onto the integration of securitypractices at a management (second wave), institu-tional (third wave) and governance (fourth wave)levels. The interweaving of private and publiccommunication networks (e.g., private compan-ies exposing themselves on the Web) has gen-erated the related security issues of cybersecur-ity (fifth wave). Here, the construction of a se-cure context over highly complex interconnectedcommunication networks (cybersecurity ERCIMNews Special Theme: Cybercrime and Privacy Is-sues 2012) should go with the help of reference or-ganisations such as the ISO/IEC (COBIT (ControlObjectives for Information and related Techno-logy) 4.1: Framework for IT Governance and Con-trol Last retrieved: June 2013; ISO/IEC 27002:2005.Information technology – Security techniques –Code of practice for information security man-agement 2005). These institutions supervise thecreation of standards whose role is to protectan organisation’s information asset in the con-text of confidentiality integrity and availability.Standards are generally largely biased, depend-ing on economical, political or simply technicalinterests. In every domain, the debate of whichstandard is better always exists. Data security isno exception (see, e.g., Solms 2005). In the par-ticular case of big data, valuable information ispotentially hidden within massive amounts ofdata. Hence, in this context, defining the cost-benefit tradeoff of securing data is a hard task.As much as there are open technical challengesfor securing data at very large-scale, there arealso strategic open questions on the overall gainof such efforts. In the context of business inform-atics, the data often originates from customerbehavior or input. Security questions thereforego beyond the technical benefits (e.g., the qual-ity of data modelling), they encompass ethicalissues, related to user rights issues related to datapreservation (e.g., including privacy, as detailedin Sect. 4.1)

3 Data Intelligence

Business intelligence is related to an accurateuse of the data collected at large scale. Themain aspects of this adequate management is theaccessibility and legibility of knowledge whereavailable, and the creation of knowledge by auto-mated or supervised processes. Figure 3 schem-atises the stages for the creation of knowledgefrom data, leading to accurate decision-makingsupport.

Data

Information

Knowledge

Filtering

Summarising

Organising

"Cleaning"

Analysis

Recognition

Synthesis

Towards Decision-making

Mining

Figure 3: Creating knowledge from data

Data is the raw material that can be collected,either as characteristics (of users, products, etc) oras traces of activities (logs, sensing, results, etc).Information is obtained by compacting data intoits coherent structures (e.g., patterns, summary).This step consists in aligning the data onto amodel. Knowledge is obtained via processingInformation with a high-level of understanding(e.g., semantic understanding). This step consistsin matching the Information with known high-level concepts.

3.1 Knowledge Management

The domain of knowledge management, via thedefinition of extensive data description schemes


offers solutions for accurate (semantic) query-based information or service access. The grail ofKnowledge Management Systems (KMS) is reas-oning and inference. From a non-redundant, butcomplete, knowledge representation, data con-tent or actor behaviour may be predicted andlinked together. However, the complexity andinduced costs of design, creation, maintenanceand compatibility of such descriptions generallyimpede their usage and development at large.

These strategies nevertheless find applicationsin well-understood, closed domains. Hence, be-sides their utility in representing knowledge viaontologies and inference models aligned withthe semantic web, KMS (Abramowicz et al. 2010;Hu et al. 2010; Jiang et al. 2009) have been ap-plied to specialised domains, including domainsrelated to business. This is the case for enterprisemodelling (Frank 2013) and cartography (Triboletand Sousa 2013) and the modelling of businessprocesses (Abramowicz et al. 2009; Sanz 2013).The reader is referred to the latter referencesfor further insights in future developments ofknowledge management in all aspects of busi-ness modelling.

3.2 Data mining and filtering

Data mining is the unsupervised (or weakly su-pervised, where weak assumption may be madeover the inner structure of the data at hand) dis-covery of recurrent or coherent patterns in thedata (Fayyad et al. 1996; Rajaraman and Ullman2012). It develops in parallel to the field of Ma-chine Learning (see, e.g., Domingos 2012) wherethe aim of the supervised process is to teach themachine specific decisions via the processing ofexamples. As such, data mining maybe coined asKnowledge Discovery (finding recurrent patternsin the data), complementing knowledge manage-ment, whereas machine learning is about know-ledge propagation (extending known decisions tounknown situations), related to the field of Pre-dictive Analysis. The flow of information evolveswith time and trends develop. Such trends are

characterised by the emergent surges of inform-ation patterns such as recurring keywords orphrases within text, or repeating events in us-age logs. A particular case of data mining, suitedto business informatics, is therefore EmergingTrend Detection (ETD) (Kontostathis et al. 2003). Itstudies flows of information along a timeline andextracts specific topic areas whose focus becomesmore important at a point of time. It should beviewed as an automated mining process, sincethe manual inspection of flows of informationat large-scale is simply not feasible. ETD is ofcrucial importance for data analysis, event pre-diction and decision-making in many areas, in-cluding business, finance, or politics. As such itis fully relevant for the analysis of big data. Byidentifying growing interests, actors of these do-mains will be able to react accordingly and evenpredict future evolution. For example, based onmining discussions over a social network, a com-pany may decide to create a new product associ-ated with a trendy product (e.g., sensitive pensfor tablets) or, on the contrary, retract a productwhose philosophy goes against current trends(e.g., large cars go against emerging “green” feel-ing). Investors may also anticipate fruitful nichesif they can detect emerging trends at an earlystage. They may also learn from the past by min-ing historical data to understand what caused thesuccess or failure of such investment or product.

ETD generally considers documents as beingaligned along a timeline and emerging trendsalso appearing and growing along that timeline(Ganesh et al. 2011). An early survey in (Kon-tostathis et al. 2003) lists and analyses systemsproposed in the late 90’s that operate on textualtechnical data such as the INSPEC database andthe IBM DB2 US Patent database. In Le et al.(2005), a technique also applying on the scientificliterature is proposed to track trendy topics usingcounts and bibliographic measures along time.Several temporal models have been proposed forthe analysis of topics over time such as the Dy-namic Topic Model (Blei and Lafferty 2006), Topics



over Time (Wang and McCallum 2006) and theTrend Analysis Model (Kawamae 2011).

The medical literature, notably with the availabil-ity of the PubMed database is a domain of interestfor ETD (Mörchen et al. 2008). Goorha and Ungar(2010) apply it to news wire articles, blogs posts,review sites and tweets, in search for interestrises in products or companies. A huge flow ofinformation is processed daily based on word andphrases counts. Leskovec et al. (2009) correlatethe appearance of given phrases in news with itsoccurrence in blogs. Similar studies have morerecently be applied on Twitter data (e.g., Wenget al. 2010).

Collaborative and hybrid recommender systems(Park et al. 2012) leverage the wisdom of thecrowd and propagate user interests across a com-munity. They can result in the emergence or fallof an item, an idea by aggregating and propagat-ing adequately consistent user judgements. Col-laborative filtering operations may be seen as alocal form of mining and trend detection withinuser interests. As such, they are also very closeto the notion of crowdsourcing (see below). Themain idea is to create a bipartite graph betweenproducts and customers where user ratings (judg-ments) are used as edge weights. Information isthen propagated along this graph to group cus-tomers and/or products and thus, predict newedge weights (i.e., the judgement of a costumerover a product).

This framework for recommendation is used inSelke and Balke (2011) to cater for the lack ofrelevant or accurate information available to cus-tomers over “experience products”. Authors demon-strate the effectiveness of their technique in thecontext of movie recommendations. This relatesthe idea of creating online and automatically itemdescriptions and therefore also relates to inform-ation retrieval. An early study on how such sys-tems may be formally evaluated is proposed inHerlocker et al. (2004).

Whereas collaborative filtering uses implicit orexplicit user judgements, sentiment analysis (a.k.a

opinion mining explores blog texts, customer re-views or comments to track the acceptance or re-jection of a product, an idea or a decision withina population (customers, voters, etc). Several ap-proaches exist, including using sentiment diction-aries to map text words to opinions or sentimentswith polarity (is/is not) (Liu 2012).

3.3 Information access: retrieval,filtering and browsing

Search and retrieval operations have installedthemselves as a base paradigm for accessing itemsfrom within a repository. They are mostly basedon the notion of a query formulation (Baeza-Yates and Ribeiro-Neto 2011). (Seidel et al. 2008),for example, demonstrates how such tools maysupport creativity in a business context.

Since precise data description is often a costlyoperation (or simply incompatible with the paceat which data is produced), in the case of systemsoperating over poorly described or non-textualdata, the idea of query-by-example has emerged(Rui et al. 1998) as a help to construct accuratequeries. Positive and negative examples are ag-gregated over intermediate search operations, inorder to form a descriptive set for the soughtitems. Examples then become the base for onlinelearning operations, so as to generalise classes ofprovided relevant and non-relevant items (Brunoet al. 2007; Wyl et al. 2011).

Browsing systems have been proposed and arealso mostly based on the definition of a searchobjective (Heesch 2008). Such systems are typic-ally oriented towards the localisation of a knowninformation, be it media copy detection or user’smental model localisation (Ferecatu and Geman2009) (see also Fig. 4). They iterate user judge-ments over appropriately-chosen sample sets ofinformation to estimate the target item the userhas in mind. This framework has been extendedby (Lofi et al. 2010) from photo search to productbrowsing for mobile e-commerce.

Adaptive Hypermedia (AH) also relies on the nav-igation paradigm for information exploration


to resolve the issue of complex query formula-tion. As accurately given in De Bra et al. (2004):

“The core problem in finding the inform-ation you want, in all the above cases, isdescribing what you want. Results fromsearch engines are often disappointing be-cause most search requests are too short andunspecific to yield good results. Once a Website with interesting information is found, itis often difficult to navigate to interestingpages only, because the site can only be nav-igated using its predefined link structure,independently of the search request thatbrought you to that site. The communityof user modelling and adaptive hypermediaoffers solutions for this problem: using in-formation gathered about the user duringthe browsing process to change the inform-ation content and link structure on-the-fly.User modelling captures the mental state ofthe user, and thus allows that knowledge tobe combined with the explicit queries (orlinks) in order to determine precisely whatthe user is looking for. To support the designof this user model-based adaptation, refer-ence models like AHAM (De Bra et al. 1999;Wu 2002) and Munich (Koch and Wirsing2002), both based on the Dexter Model byHalasz and Schwartz (1994), have been in-troduced in an attempt to standardise andunify the design of adaptive hypermediaapplications, used mostly in isolated inform-ation spaces such as an online course, anelectronic shopping site, an online museum,etc”.

In Brusilovsky (2001), a taxonomy of AH tech-nologies is further presented. The taxonomy isanalysed in detail in Stash (2007), along with anextensive review of AH systems.

The above involves the notion of user modellingand a comprehensive review on personalisationresearch in e-commerce is presented in Adolphsand Winkelmann (2010).

Information filtering comes as a helper solutionfor the interactive formulation of search queries.Rules are defined over product characteristics, inorder to define the class of the sought items as theintersection of solution sets for the rules. Rulesare generally based on information facets. Facetsare orthogonal, mutually exclusive dimensionsof the data whose range is quantised in relevantintervals (Hearst 2008). Facets may be determ-ined from the data model itself by highlightingimportant characteristics of the data. In explor-atory conditions however, i.e., when the data isnot fully understood, it may be interesting tolet facets emerge automatically or interactivelyfor providing interpretation of its organisationand to facilitate its exploration (Zwol and Sigur-björnsson 2010). Several routes may be taken toautomatically determine data facets. They all con-sist in using the data or a representative samplein a mining process to identify a reduced set oforthogonal projection operators whereby everydata item is identified by its set of projections.

Faceted search is extensively used over e-com-merce sites when products bear inherent ortho-gonal characteristics. For example, this is thecase for real estate commerce with facets such asproduct type, surface, region, price range, ...

3.4 Visual analytics

The above tools are used to make sense of thedata itself, using the intrinsic data content or theusage context of the data. Visual analytics refersto “the science of analytical reasoning facilitatedby interactive visual interfaces” (Heer and Schnei-dermann 2012; Keim et al. 2010; Thomas andCook 2006) and creates a link between content-based data mining and interactive data explora-tion techniques, as described above.

Visual Analytics supports the user in exploringthe data and to interactively guide the system tofind a formal solution that matches an intuitivesolution (the mental model) to the problem (seeFig. 4).



Data

Use case

(problem)

Data Analysis Tool

(support)

Mental ModelInteractive

Visualisation

User

(decision)

Figure 4: The process of Visual Analytics where the useris matching a mental model of the solution with theknowledge inferred from the data (see also Fig. 3)

In Zhang et al. (2012), a review of commercial sys-tems for Visual Analytics, to support facing thisbig data era is proposed in the context of Busi-ness Intelligence. Various use cases are explored(inc. medical, microblogging) and performanceover factors such as scalability and effectivenessfor supporting decision-making are given. Au-thors then issue a number of future challengesrelated to effective large-scale data analysis.

One of the key parameters in interactive dataanalysis is to offer proper user interfaces and toadequately leverage the potential of user inter-action, seen as a source of semantic knowledgeinto the system (Morrison et al. 2012). The roleof users and user groups is studied in the nextsections.

4 From the user to the community

While an accurate use of the data is fundamentalto the decision process, the ultimate actor in theprocess remains the user. There are many user-related issues technologies should take care of.

As much as data should be secured, the privacyof a user should be guaranteed. This will allowthe user to act freely in the environment s-he isconfronted with. Consequently, a consistent andreliable user behaviour may lead to trust and highreputation that may be used in many contextsfor information access and recommendation.

Thanks to ever-developing communication me-dia, users may also group into communities andform social groups. The emergence and identi-fication of such social networks allows the ana-lysis to move from the individual to the proto-typical user-community s-he belongs to. Thiselectronic crowd represents a task-force and amass of semantic knowledge that crowdsourcingefforts aim at capturing.

4.1 User privacyUser privacy (Danezis and Gürses 2010; ERCIMNews Special Theme: Cybercrime and PrivacyIssues 2012; Hansen et al. 2008) is directly relatedto data security. The relationship between datasecurity and user privacy is established via “guar-anteeing privacy by securing access to privatedata”.

User privacy can be understood as a two-foldconcept, ethical and technological. Ethics shouldprevent the usage of user data to infer specificuser needs and thus make that user fragile overcommunication networks. User data should bestudied statistically and anonymously so that itreturns to the user as a member of one user class,not as an individual. Many more ethical aspectsshould be defined in parallel of the advent ofbig data (Davis 2012). This is the role of govern-mental or not-for-profit independent organisa-tions (e.g., the UN World Trade Organisation) tocounter the temptation of inadequate usage ofthis data from large Internet companies, eventhough it is known that individuals value theirprivacy but tend to give it up easily as customers(Pogue 2011).

Technological solutions should ensure that theuser data and behavior (e.g., mirrored into us-age logs) remain private and are not accessible


in their raw form to anyone. Anonymity maybe a solution to privacy (Edman and Yener 2009),but again, this approach may not be feasible any-more, as soon as the individual becomes a cus-tomer or a user of social networks (De Cristofaroet al. 2012; Fung et al. 2010).

Also related to privacy is the possibility for se-cure authentication (Poller et al. 2012), prevent-ing identity spoofing. These fields, associated todigital forensic and secured biometrics, directlyrelate to the notion of trust over communicationnetworks.

4.2 Modelling trust

Trust is a social notion that an individual or agroup (persons or organisation) develops overtime and along experience. It measures the beliefthat the actions of an individual or a group maybe predicted (e.g., from social knowledge of theindividual or group) and stay within the limits ofa predefined frame. Trust is closely related to thenotion of reputation (Castelfranchi and Falcone1998; Pinyol and Sabater-Mir 2013). It is opposedto the adverse behaviour of cheating via fraudand attacks (Hoffman et al. 2009). As such, theestimation of trust and reputation represents theestimation of a risk for the environment wherethe individual or group in question is active.

Models for trust and reputation over communic-ation networks such as the Internet have beenproposed with essentially two approaches. Thegame theory approach formalises a competitioncontext where the objective is to maximise payoffwith minimising risks. Trust estimation thereforerelies on associated risk-minimisation tools. Thecognitive approach accounts for elements suchas beliefs, goals, desires and intentions. As suchthe resulting trust models bear as much valuein their result than in their capability to explainthe result. A thorough review of these modelsand their classification is proposed in Pinyol andSabater-Mir (2013). These models are importantto estimate the value of user interaction in sys-tems such as recommender systems (Maida et al.2012) (see also Sect. 3.3 above).

4.3 Social Network analysis

The constitution of social communities andgroups of interest have allowed to move fromthe individual perspective to mass-address forbusiness (essentially for push-based advertise-ment, the main revenue model for the Web). Thestudy of social networks is therefore essentialto structure the potential of such communities,including via the detection of key network fea-tures such as connectivity and influential nodes(Gomez-Rodriguez et al. 2012; Sun et al. 2013).

In relation to adaptive hypermedia and recom-mendation systems, where it is the study of userinteraction that leads to recommendation, thestudy of social media (media hyperlinked in so-cial networks) may allow the inference of re-commendation (friends over Facebook or connec-tions over LinkedIn) (Backstrom and Leskovec2011). One of the difficulties here is the scale atwhich algorithms should operate. A compensat-ing advantage of human-structured networks istheir reputed low diameter (originally valued to 6(Schnettler 2009), but said to be reduced to 4 oversocial networks) enabling local computations.

4.4 Social labor: Human Computationand crowdsourcing

There is a large labor potential to leverage overthe Internet. This is known as Human Com-putation (Ahn 2005; Quinn and Bederson 2011)and also relates to crowdsourcing (Jones 2013).This strategy is, for example, used to help di-gitising characters via the ReCAPTCHA (Com-pletely Automated Public Turing Test To TellComputers and Humans Apart) system. Here,the trust in the user is evaluated by presentinga problem with a known answer. The answer toan unknown problem proposed simultaneouslyis then used as a statistical clue towards theright solution of this latter problem. In general,these tools, along with the Games With a Purpose(GWAP6), use the fact that human capabilitiesto perform (visual) pattern recognition surpass

6Games With a Purpose: http://www.gwap.com

http://www.gwap.com



by far that of an automated process, with theincentive of fun or commercial advantage. Re-commender systems may also be seen as a formof crowdsourcing in that they seamlessly federateuser judgements to create semantic informationabout items, products or services.

The impressive performance of such collaborat-ive systems demonstrate the potential of labor tobe federated over the internet. Another way offederating the crowd as a workforce is the useof digital labor (Larson et al. 2012). For example,the Amazon Mechanical Turk mediates betweenjob requesters and workers. A requester createsa HIT (Human Intelligence Task) and proposesa reward for it. This HIT generally consists ofa short but repetitive task such as asserting thepresence of an object in an image. The trust intoworkers’ competences may be evaluated by ini-tial trials and a reputation system is active forboth workers and requesters.

Eventually, if enough workers act on a simpletask, this workforce constitutes a parallel pro-cessing machine (e.g., the click-farms to cheatInternet ads) and software APIs have even beendeveloped to make that process fully transparent.

4.5 Social knowledge : Folksonomies

While human labor may be organised over theInternet, there are also several initiatives to feder-ate human knowledge, following the “Wisdom ofthe Crowd” paradigm (Surowiecki 2004). Beyondthe ever-growing Wikipedia and its collaborat-ive edition model, including trust and reputationmechanisms, the combination of the semanticweb and Web 2.0 for social behaviour enablesthe gathering of a social knowledge, known asfolksonomy (Lohmann and Díaz 2012). This know-ledge is made accessible accessible to machinesvia semantic web technologies and also offers agreat potential for the development of adaptiveor human-tailored business services.

5 Discussion and conclusion

Modern communication networks have fueledthe creation of massive volumes of data. In thispaper, we have discussed how this data may be-come an asset for business activities. This thor-ough overview of the information landscape, aug-mented with a large number of key referencesaims at providing a faithful picture and guidelinefor the practitioner who wants to attack the prob-lem of data management and analysis in a busi-ness context. We highlighted and exemplified thepotential benefits of data analysis but also thecomplexity and issues related to this task. Weadvocated for considering in parallel the dataand the user viewpoints. Both perspectives sharecommonalities in their structure and analysis.The first being that most of the data originatesfrom the users and that the users will then bemodeled (in their behavior) via the analysis ofdata. Further, as much as data may be seen atdifferent scales, user and user communities maybe modeled at different scales. There is thereforemuch to gain in keeping this relationship alivewhen exploring and exploiting the data.

In this era of big data, large-scale data analysisbecomes a strategic field of development. Thepromise of a reasoning machine by the field ofartificial intelligence in the 1960’s has been re-placed by the statistical crunching of massivedata with the side effect of smoothing out inter-esting details. The original three Vs of big dataimpose shallow processing for scalability. It isstill an open challenge to design scalable processto filter (project or denoise, however a volumereduction strategy may be based on) the data tolower volumes and enable more effective ana-lysis. In parallel, distributed infrastructures ac-commodating hierarchical processing of the datamay help finding the essence of information andfocus on these sparse interesting needles in thedata haystack. It is also a commonplace that thepotential of big data for business profitability ismore an intuition than a frequent reality7. Hence,

7https://www.facebook.com/dan.ariely/posts/904383595868

https://www.facebook.com/dan.ariely/posts/904383595868

https://www.facebook.com/dan.ariely/posts/904383595868


not only effective processing and efficient infra-structures are needed but also the right analysismodels are still lacking.

The ubiquity of communication platforms andnetworks have shaped the culture and raised newconcerns and approaches towards privacy andsecurity issues. Further deployments are likelyto be guided by the further integration of con-nected hardware into our everyday lives. Thenews ways of interacting with the data that thesedevices create may be exploited for further aug-menting the ubiquity and usefulness of the datafor the customer. In turn, these powerful sensingdevices will enable companies to tailor recom-mendation and targeting systems even further.Where the original Web was about simple data,the Web 2.0 about people and their relationships,the emergent web of things (suggested as “Web3.0”) proposes to connect “Anything, Anywhere,Anytime”. Objects will be part of the communica-tion process and their usage, location, proximity,etc will be tracked for better user behavior under-standing. Extending the concept of Tangible UserInterface, devices may symbol any data or virtualentity and be moved, exchanged, or combined asany object can be, with an associated impact inthe virtual world.

Affective computing (Picard 2000), assessing useremotions via physiological sensors will also allowproviders to penetrate even further into users’wishes. The trade-off between privacy and utilityis thus very likely to continue evolving.

6 Acknowledgments

This work has been partly funded by Swiss Na-tional Science Foundation (SNSF) via the SwissNCCR(IM)2 and the Swiss State Secretariat forEducation and Research (SER) supporting ICTCOST Action IC1002 “Multilingual and Multifa-ceted Interactive Information Access” (MUMIA).

References

Abramowicz W., Haniewicz K., Kaczmarek M.,Zyskowski D. (2009) Semantic Modelling ofCollaborative Business Processes. In: KusiakA., goo Lee S. (eds.) eKNOW. IEEE Comp.Soc., pp. 116–122

Abramowicz W., Fensel D., Frank U. (2010) Se-mantics and Web 2.0 Technologies to Sup-port Business Process Management. In: Busi-ness & Information Systems Engineering 2(1),pp. 1–2

Adolphs C., Winkelmann A. (2010) A rigorousliterature review on personalization researchin e-commerce (2000-2008). In: Journal of Elec-tronic Commerce Research 11(4), pp. 326–341

von Ahn L (2005) Human Computation. PhDthesis, Carnegie Mellon University Last Ac-cess: (UMI Order Number: AAI3205378)

Alnatheer M., Chan T., Nelson K. (2012) Under-standing And Measuring Information Secur-ity Culture. In: Pacific Asia Conference onInformation Systems (PACIS2012) Proceed-ings. Ho Chi Minh City, Vietnam

Backstrom L., Leskovec J. (2011) Supervised ran-dom walks: predicting and recommendinglinks in social networks. In: Poceedings ofthe WSDM’2011 Conference, pp. 635–644

Baeza-Yates R., Ribeiro-Neto B. (2011) ModernInformation Retrieval: the concepts and tech-nology behind search, 2nd. Add. Wesley

Big Data: Science in the Petabyte Era. Nature,vol. 455, num. 7209. Last Access: (special is-sue)

Blei D. M., Lafferty J. D. (2006) Dynamic topicmodels. In: Proceedings of the 23rd interna-tional conference on Machine learning. ICML’06. Pittsburgh, Pennsylvania, pp. 113–120

Bruno E., Kludas J., Marchand-Maillet S. (2007)CombiningMultimodal Preferences for Multi-media Information Retrieval. In: Proceedingsof the international workshop on Workshopon multimedia information retrieval

Brusilovsky P. (2001) Adaptive Hypermedia. In:User Modelling and User-Adapted Interac-tion 11, pp. 87–110



Castelfranchi C., Falcone R. (1998) Social Trust.In: Proceedings of the first workshop on de-ception, fraud and trust in agent societies.Minneapolis, USA

COBIT (Control Objectives for Information andrelated Technology) 4.1: Framework for ITGovernance and Control Institute, Informa-tion Systems Audit and Control Association(ISACA) http://www.isaca.org

Conklin W. (2007) Barriers to Adoption ofe-Government. In: System Sciences, 2007.HICSS 2007, pp. 98–98

Danezis G., Gürses S. (2010) A critical reviewof 10 years of privacy technology. In: Surveil-lance Cultures: A Global Surveillance Soci-ety? UK

Davis K. (2012) Ethics of Big Data. O’Reilly Me-dia

De Bra P., Houben G.-J., Wu H. (1999) AHAM:A Dexter-based Reference Model for Adapt-ive Hypermedia. In: Proceedings of the 10thACM conference on Hypertext and Hyperme-dia. Darmstadt

De Bra P., Aroyo L., Chepegin V. (2004) The NextBig Thing: Adaptive Web-Based Systems. In:Journal of Digital Information 5(1)

De Cristofaro E., Soriente C., Tsudik G., Willi-ams A. (2012) Hummingbird: Privacy at theTime of Twitter. In: Security and Privacy (SP),2012 IEEE Symposium on, pp. 285–299

Dean J., Ghemawat S. (2004) MapReduce: Sim-plified Data Processing on Large Clusters. In:OSDI’04: Sixth Symposium on Operating Sys-tem Design and Implementation. San Fran-cisco, CA

Diesburg S. M., Wang A.-I. A. (Dec. 2010) A sur-vey of confidential data storage and deletionmethods. In: ACM Comput. Surv. 43(1), 2:1–2:37

Domingos P. (Oct. 2012) A few useful things toknow about machine learning. In: Commun.ACM 55(10), pp. 78–87

Edman M., Yener B. (Dec. 2009) On anonym-ity in an electronic society: A survey of an-onymous communication systems. In: ACMComput. Surv. 42(1), 5:1–5:35

ERCIM News Special Theme: Big Data. 89ERCIM News Special Theme: Cybercrime and

Privacy Issues. 90Fayyad U., Piatetsky-Shapiro P., Smyth P. (1996)

Knowledge Discovery and Data Mining: To-wards a unified framework. In: Proceedingsof the ACM SIG KDD Conference

Fayyad U. (2012) Big Data Analytics: Applica-tions and Opportunities in On-line Predict-ive Modeling. In: Proceedings of the ACMSIGKDD Conference. (material available on-line – June 2013)

Ferecatu M., Geman D. (2009) A statistical frame-work for image category search from a men-tal picture. In: IEEE Transactions on Pat-tern Analysis and Machine Intelligence 31(6),pp. 1087–1101

Frank U. (2013) Enterprise Modelling: the nextsteps. In: IEEE Conference on Business In-formatics (IEEE–CBI 2013). Vienna, Austria

Fung B. C. M., Wang K., Chen R., Yu P. S.(June 2010) Privacy-preserving data publish-ing: A survey of recent developments. In:ACM Comput. Surv. 42(4), 14:1–14:53

Ganesh M. S., Reddy C. P., N.Manikandan, Ven-kata D. P. (2011) TDPA: Trend Detection andPredictive Analytics. In: International Journalon Computer Science and Engineering 3(3)

Gomez-Rodriguez M., Leskovec J., Krause A.(Feb. 2012) Inferring Networks of Diffusionand Influence. In: ACM Trans. Knowl. Discov.Data 5(4), 21:1–21:37

Goorha S., Ungar L. (2010) Discovery of signific-ant emerging trends. In: Proceedings of the16th ACM SIGKDD international conferenceon Knowledge discovery and data mining.Washington, DC, USA, pp. 57–64

Greene G., D’Arcy J. (2010) Security Culture andthe Employee-Organization Relationship inIS Security Compliance. In: Proc. of the 5thAnnual Symposium on Information Assur-ance. New York, USA

Halasz F., Schwartz M. (1994) The Dexter Hyper-text Reference Model. In: Communications ofthe ACM 37(2), pp. 30–39

Hansen M., Schwartz A., Cooper A. (2008) Pri-


vacy and Identity Management. In: SecurityPrivacy, IEEE 6(2), pp. 38–45

Hearst M. A. (2008) UIs for Faceted Navigation:Recent Advances and Remaining Open Prob-lems. In: Workshop on Computer Interactionand Information Retrieval, HCIR. Redmond,WA

Heer J., Schneidermann B. (2012) Interactive Dy-namics for Visual Analytics. In: Communica-tion of the ACM 55(4)

Heesch D (2008) A survey of browsing modelsfor content based image retrieval. In: Multi-media Tools and Applications 40 (2)

Herlocker J. L., Konstan J. A., Terveen L. G.,Riedl J. T. (Jan. 2004) Evaluating collaborat-ive filtering recommender systems. In: ACMTrans. Inf. Syst. 22(1), pp. 5–53

Hoffman K., Zage D., Nita-Rotaru C. (Dec. 2009)A survey of attack and defense techniquesfor reputation systems. In: ACM ComputerSurveys 42(1), 1:1–1:31

Hu S., Wan L., Zeng R. (2010) Web2.0-based En-terprise Knowledge Management Model. In:Information Management, Innovation Man-agement and Industrial Engineering (ICIII),2010 International Conference on Vol. 4,pp. 476–480

ISO/IEC 27002:2005. Information technology –Security techniques – Code of practice forinformation security management (update ofISO/IEC 17799) International Standard Organ-isation

Jiang H., Liu C., Cui Z. (2009) Research on Know-ledge Management System in Enterprise. In:Computational Intelligence and Software En-gineering, CiSE 2009. International Confer-ence on, pp. 1–4

Jones G. J. F. (2013) An introduction to crowd-sourcing for language and multimedia tech-nology research. In: Proceedings of the2012 international conference on Informa-tion Retrieval Meets Information Visualiz-ation. PROMISE’12. Springer-Verlag, Zinal,Switzerland, pp. 132–154

Kawamae N. (2011) Trend analysis model: trendconsists of temporal words, topics, and

timestamps. In: Proceedings of the Forth In-ternational Conference on Web Search andWeb Data Mining, WSDM 2011, Hong Kong,China, pp. 317–326

Keim D., Kohlhammer J., Ellis G., Mansmann F.(eds.) Mastering the Information Age – Solv-ing Problems with Visual Analytics. Euro-graphic Digital Library

Koch N., Wirsing M. (2002) The Munich Ref-erence Model for Adaptive Hypermedia Ap-plications. In: 2nd International Conferenceon Adaptive Hypermedia and Adaptive Web-Based Systems, pp. 213–222

Kontostathis A., Galitsky L., Roy S., PottengerW.M., Phelps D. (2003) A survey of ETD in Tex-tual Data Mining. In: Berry M. (ed.) A Com-prehensive Survey of Text Mining. Springer

Larson M., Soleymani M., Eskevich M., Serdy-ukov P., Jones G. J. (2012) The Communityand the Crowd: Developing large-scale datacollections for multimedia benchmarking. In:IEEE Multimedia, Special Issue on Large-Scale Multimedia Data Collections

Le M.-H., Ho T.-B., Nakamori Y (2005) Detect-ing Emerging Trends from Scientific Corpora.In: International Journal of Knowledge andSystems Sciences 2(2)

Leskovec J., Backstrom L., Kleinberg J. (2009)Meme-tracking and the dynamics of thenews cycle. In: Proceedings of the 15th ACMSIGKDD international conference on Know-ledge discovery and data mining. KDD 2009.Paris, France, pp. 497–506

Liu B. (2012) Sentiment Analysis and OpinionMining. Synthesis Lectures on Human Lan-guage Technologies 1 Vol. 5. Morgan & Clay-pool

Lofi C., Nieke C., Balke W.-T. (2010) MobileProduct Browsing Using Bayesian Retrieval.In: 12th Conference on Commerce and En-terprise Computing (CEC 2010). Shanghai,China, pp. 96–103

Lohmann S., Díaz P. (2012) Representing andvisualizing folksonomies as graphs: a ref-erence model. In: Proceedings of the Inter-national Working Conference on Advanced



Visual Interfaces (AVI’12). ACM, Capri Island,Italy, pp. 729–732

Maida M., Maier K., Obwegeser N., Stix V. (2012)A Multidimensional Model of Trust in Recom-mender Systems. In: EC-Web, pp. 212–219

Manyika J., Chui M., Brown B., Bughin J., DobbsR., Roxburgh C., Byers A. H. (2011) Big data:The next frontier for innovation, competition,and productivity. Mc Kinsey Global Institute

Mohamed H., Marchand-Maillet S. (2012) Dis-tributed media indexing based on MPI andMapReduce. In: Multimedia Tools and Applic-ations, pp. 1–25

Mörchen F., Dejori M., Fradkin D., Etienne J.,Wachmann B., Bundschus M. (2008) Anticip-ating annotations and emerging trends in bio-medical literature. In: Proceedings of the 14thACM SIGKDD international conference onKnowledge discovery and data mining. LasVegas, Nevada, USA, pp. 954–962

Morrison D., Tsikrika T., Hollink V., de VriesandE. Bruno A. P., Marchand-Maillet S. (2012)Topic modelling of clickthrough data in im-age search. In: Multimedia Tools and Applic-ations, pp. 1–23

Mukherjee A., Liu B., Glance N. (2012) SpottingFake Reviewer Groups in Consumer Reviews.In: Proceedings of the 21st International Con-ference onWorld WideWeb. WWW’12. Lyon,France, pp. 191–200

Park D. H., Kim H. K., Choi I. Y., Kim J. K. (2012)A Literature Review and Classification of Re-commender Systems Research. In: Expert Sys-tems with Applications (in press)

Picard R. W. (2000) Affective Computing. MITPress

Pinyol I., Sabater-Mir J. (2013) Computationaltrust and reputation models for open multi-agent systems: a review. In: Artificial Intelli-gence Review 40(1), pp. 1–25

Place I., Hyslop D. (1982) Records management:controlling business information. Reston Pub.Co.

Pogue D. (2011) Don’t Worry about Who’swatching. In: scientific American 304(1), p. 32

Poller A., Waldmann U., Vowe S., Turpe S. (2012)

Electronic Identity Cards for User Authen-tication – Promise and Practice. In: SecurityPrivacy, IEEE 10(1), pp. 46–54

Quinn A. J., Bederson B. B. (2011) Human com-putation: a survey and taxonomy of a grow-ing field. In: Proceedings of the SIGCHI Con-ference on Human Factors in Computing Sys-tems. CHI ’11. ACM, Vancouver, BC, Canada,pp. 1403–1412

Rajaraman A., Ullman J. D. (2012) MiningMassive Datasets. Cambridge UniversityPress

Rui Y., Huang T. S., Ortega M., Mehrotra S.(1998) Relevance Feedback: A Power Tool inInteractive Content-Based Image Retrieval.In: IEEE Trans. on Circuits and Systems forVideo Technology 8(5)

Ruighaver A., Maynard S., Chang S. (2007) Or-ganisational security culture: Extending theend-user perspective. In: Computers and Se-curity 26(1)

Sanz J. L. C. (2013) Enabling Customer Ex-perience and Front-office Transformationthrough Business Process Engineering. In:IEEE Conference on Business Informatics(IEEE-CBI 2013). Vienna, Austria

Schnettler S. (2009) A structured overview of50 years of small-world research. In: SocialNetworks 31(3), pp. 165 –178

Seidel S., Muller-Wienbergen F. M., RosemannM., Becker J. (2008) A Conceptual Frame-work for Information Retrieval to SupportCreativity in Business Processes. In: Proceed-ings 16th EuropeanConference on Informa-tion Systems. Galway, Ireland

Selke J., Balke W.-T. (2011) Turning ExperienceProducts into Search Products: Making UserFeedback Count. In: 13th IEEE Conf. on Com-merce and Enterprise Computing (CEC 2011).Luxembourg

von Solms B. (2005) Information Security gov-ernance: COBIT or ISO 17799 or both? In:Computers and Security 24(2)

von Solms B. (2006) Information security – theFourth Wave. In: Computer and Security25(3), pp. 165–168


von Solms B. (2010) The 5 Waves of Informa-tion Security – From Kristian Beckman to thePresent. In: Rannenberg K., Varadharajan V.,Weber C. (eds.) Security and Privacy – SilverLinings in the Cloud. IFIP Advances in In-formation and Communication TechnologyVol. 330. Springer Berlin Heidelberg, pp. 1–8

Stash N. (2007) Incorporating Cognit-ive/Learning Styles in a General-PurposeAdaptive Hypermedia System. PhD thesis,Eindhoven University of Technology, TheNetherlands

Sun K., Morrison D., Bruno E., Marchand-Maillet S. (2013) Learning RepresentativeNodes in Social Networks. In: 17th Pacific-Asia Conference on Knowledge Discoveryand Data Mining. Gold Coast, AU

Surowiecki J. (2004) The wisdom of crowds:Why the many are smarter than the fewand how collective wisdom shapes business,economies, societies and nations. New York:Doubleday

Thomas J., Cook K. (2006) A Visual AnalyticsAgenda. In: IEEE Computer Graphics and Ap-plications 26(1), pp. 10–13

Tribolet J., Sousa P. (2013) Enterprise Gov-ernance and Cartography. In: IEEE Conf. onBusiness Informatics (IEEE-CBI 2013). Vienna

Vossen G. (2013) Big data as the new enabler inbusiness and other intelligence. In: VietnamJournal of Computer Science 1(1), pp. 1–12

Wang J.-F. (2009) E-government Security Man-agement: Key Factors and Countermeasure.In: Information Assurance and Security, 2009.IAS09. Fifth International Conference onVol. 2, pp. 483–486

Wang X., McCallum A. (2006) Topics over time:a non-Markov continuous-time model of top-ical trends. In: Proceedings of the 12th ACMSIGKDD international conference on Know-ledge discovery and data mining. KDD ’06.Philadelphia, PA, USA, pp. 424–433

Weng J., Lim E.-P., Jiang J., He Q. (2010) Twitter-Rank: finding topic-sensitive influential twit-terers. In: Proceedings of the third ACM in-ternational conference on Web search and

data mining. Proceedings of the third ACMinternational conference on Web search anddata mining WSDM ’10. New York, New York,USA, pp. 261–270

Wu H (2002) A Reference Architecture for Ad-aptive Hypermedia Applications". PhD thesis,Eindhoven University of Technology

von Wyl M., Mohamed H., Bruno E., Marchand-Maillet S. (2011) A parallel cross-modalsearch engine over large-scale multimediacollections with interactive relevance feed-back. In: Proceedings of the 1st ACM Interna-tional Conference on Multimedia Retrieval.ICMR ’11. ACM, Trento, Italy, 73:1–73:2

Zhang L., Stoffel A., Behrisch M., Mittelstadt S.,Schreck T., Pompl R., Weber S., Last H., KeimD. (2012) Visual analytics for the Big Dataera – A comparative review of state-of-the-art commercial systems. In: Visual AnalyticsScience and Technology (VAST), 2012 IEEEConference on, pp. 173–182

van Zwol R., Sigurbjörnsson B. (2010) Facetedexploration of image search results. In: Pro-ceedings of the 19th international conferenceon World Wide Web (WWW’2010)

Stéphane Marchand-Maillet

Department of Computer ScienceCentre Universitaire Informatique (CUI)University of [email protected]

Birgit Hofreiter

Electronic Commerce GroupInstitute of Software Technology and InteractiveSystemsVienna University of [email protected]


106 Thomas Setzer

Thomas Setzer

Data-Driven Decisions in Service Engineering andManagement

Today, the frontier for using data to make business decisions has shifted, and high-performing servicecompanies are building their competitive strategies around data-driven insights that produce impressivebusiness results. In principle, the ever-growing amount of available data would allow for deriving increasinglyprecise forecasts and optimised input for planning and decision models. However, the complexity resultingfrom considering large volumes of high-dimensional, fine-grained, and noisy data in mathematical modelsleads to the fact that dependencies and developments are not found, algorithms do not scale, and traditionalstatistics as well as data-mining techniques collapse because of the well-known curse of dimensionality.Hence, in order to make big data actionable, the intelligent reduction of vast amounts of data to problem-relevant features is necessary and advances are required at the intersection of economic theories, servicemanagement, dimensionality reduction, advanced analytics, robust prediction, and computational methods tosolve managerial decisions and planning problems.

1 Introduction

Increasingly automated data capturing, the ubi-quity of sensors, the spread of smart phones, andthe penetration of life by social media leads toenormous and ever growing amounts of data.Novel technological advances in analytics andscalable data management promise to facilitatethe capturing, storage, searching, sharing, analys-ing, and visualisation of relationships and trendshidden in large, high-dimensional data sets.

While, traditionally, scientists in areas such asmeteorology, genomics, physic simulations, orenvironmental research were primarily faced withthe challenges of exploring large, very high-di-mensional data sets, today such challenges alsoaffect areas like business informatics. In par-ticular service design and management need toprocess data in order to spot business trends, de-termine and anticipate bottlenecks and quality ofservice, or prevent customer churn by identify-ing churn risk and triggering appropriate actions,to name only a few tasks. In general, enterprisesthat can use their data quickly and correctly can

gain efficiency through data-driven decisions, an-ticipatory action and accelerated service supportand delivery processes. As an example, thosecompanies can utilise knowledge extracted frompast customer behaviours to better understandcustomers in order to better convince them withsmart, individualised offers and services.

1.1 Service Management

Traditionally, the aim of service managementis to optimise service-intensive supply chains,which are typically much more complex than thesupply chains of typical goods. Those requiretighter integration with field service and thirdparties and must also accommodate inconsistentand uncertain demand by establishing more in-tegrated and more robust information flows. Inaddition, most processes must be coordinatedacross numerous service locations. Interestingly,among typical manufacturers, after-sale services(support, repair, maintenance, etc.) comprise lessthan 20% of revenue. Among the most successfulcompanies, those same activities on average gen-erate more than 50 percent of their total profits

Enterprise Modelling and Information Systems ArchitecturesVol. 9, No. 1, June 2014Data-Driven Decisions in Service Engineering and Management 107

(Accenture 2006). This is one of many observa-tions indicating that a profound understandingof customers and business partners and establish-ing high-quality service and information man-agement is of crucial importance.

However, today enterprises provide an increas-ing number of services in an automated or semi-automated fashion by means of information tech-nology (IT services), where customer behaviourand experience can only be ‘observed’ by track-ing what a customer is doing, in particular howhe uses one or more services over time. Providerseven of IT-only services can no longer afford tofocus on technology and their internal organisa-tion, but need to consider the quality of the ser-vices they provide and focus on the relationshipwith customers. IT service management (ITSM)refers to the implementation and managementof high quality IT services that meet the needsof customers. ITSM is performed by IT serviceproviders through an appropriate mix of people,process and information technology (Office ofGovernment Commerce (OGC) 2009).

Unfortunately, in particular with IT services, pro-viders typically do not receive regular direct cus-tomer feedback that is required for marketing,further service improvements, and service innov-ation. However, there is an ever-growing amountof information how a customer uses a services(e.g., sensors of a rental car, log files of a Web-shop, browsing behaviour in on-line manuals,etc.), and these datasets can be analysed to get‘implicit’ feedback as described for example inChoi and Ahn (2009).

1.2 Advanced Analytics

In fact, today’s service enterprises have moredata at hand about their markets, customers, andrivals than ever before. Analysing those vastamounts of historical and current data in an auto-nomic or semi-autonomic fashion allows for pre-dicting service demand and usage, customer be-haviour, and market dynamics. In addition, it

allows for identifying novelty patterns in cus-tomer behaviour and improving short and long-term performance of enterprise business systems,which is vital for running a competitive servicecompany.

In ‘Competing on Analytics: The New Science ofWinning’, Davenport and Harris (2006) argue thatthe frontier for using data to make business de-cisions has shifted. Many high-performing com-panies are building their competitive strategiesaround data-driven insights that generate im-pressive business results. Those companies useadvanced analytical procedures, sophisticatedquantitative and statistical analysis and predict-ive modelling. Examples of analytics are the us-age of novel tools to determine the most profit-able customers and offer them the right price, toaccelerate product innovation, to optimise andintegrate supply chains, and to identify the majordrivers of financial performance. Many examplesfrom organisations such as Amazon, Barclay’s,Capital One, Harrah’s, and Procter & Gamble arepresented, showing how to leverage analytics todrive business. However, various potential defin-itions for advanced analytics exist. Typically, the‘advanced’ indicates quantitative, predictive orprescriptive models as described later in this pa-per.

1.3 Big Data Analytics

Over the last two years, the term Big data ispropagated by major companies offering inform-ation management software such as Intel1, SAP2,or IBM3, and has become more and more a syn-onym for data analysis and advanced analyt-ics. For many SMEs and also for larger com-panies, this is in some sense counter-productiveas nowadays enterprises collect massive amounts

1http://www.intel.de/content/www/de/de/big-data/big-data-analytics-turning-big-data-into-intelligence.html

2http://www54.sap.com/pc/tech/in-memory-computing/hana/software/analytics/big-data.html

3http://www-01.ibm.com/software/data/infosphere/hadoop/what-is-big-data-analytics.html

http://www.intel.de/content/www/de/de/big-data/big-data-analytics-turning-big-data-into-intelligence.html



http://www54.sap.com/pc/tech/in-memory-computing/hana/software/analytics/big-data.html



http://www-01.ibm.com/software/data/infosphere/ hadoop/what-is-big-data-analytics.html

http://www-01.ibm.com/software/data/infosphere/ hadoop/what-is-big-data-analytics.html


108 Thomas Setzer

of various metrics, such as historical sensor, mon-itoring, and customer usage data, hoping thatthe data will turn out to be useful one day forprediction and optimisation.

Accordingly, as Big data analytics is now a pop-ular topic for management, many informationmanagement companies offer tools and solutionsto extract and project relationships between avast amount of high-dimensional data vectors(structured, semi-structured, or unstructuredones), and to process, reduce, correlate and inter-pret data in a much more flexible fashion com-pared to traditional database management andbusiness intelligence systems.

Over the last years, enterprises such as SoftwareAG, Oracle, IBM, Microsoft, SAP, EMC, and HPhave spent more than $15 billion on softwarefirms only specialising in data management andanalytics. Since the last three years, this industrywas worth more than 100 billion US-dollars andwas growing at around 10 percent a year: abouttwice as fast as the software business in general(The Economist 2010).

1.4 The Curse of Dimensionality

While in principle the vast and ever-growingsets of available data would allow for derivingincreasingly precise predictions and optimisedplanning and decision models, the complexityresulting from the consideration of large volumesof multivariate, fine-grained, often noisy and in-complete data leads to the fact that relationshipswithin the data are not found, algorithms do notscale, and traditional statistics as well as data-mining techniques collapse because of the well-known curse of dimensionality (nowadays alsocalled the curse of big data) (Bellman 1961; Leeand Verleysen 2007).

Despite these dimensionality-intrinsic problems,biases in how data are collected, a lack of context,gaps in what’s gathered, artefacts of how data areprocessed and the overall cognitive biases thatlead even experienced researchers to determinenon-existing patterns (and vice versa) shows that

even if a company has Big Data, making use ofsuch data typically not only requires appropriatetools but also data scientists with expertise andknow-how, hacking-skills, domain knowledge,and deep mathematical and data managementskills; unfortunately, as of yet data scientists ofthat sort are still a very scarce human resource(Davenport and Patil 2012).

The result is that – in practice – data are oftencollected and then ignored or aggregated in aproblem-agnostic fashion, and finally for mostproblems rather simple and conservative solutionheuristics are applied by rules of thumb or usingcoarsened data. The authors of this article are notaware of many companies besides the financialinstitutions and telecommunications companiesthat make excessive use of their collected data;however, most enterprises spend an increasingamount of money and effort in monitoring sys-tems and data collection. That is also the out-come of numerous studies and expert interviewsconducted and summarised by Ross et al. (2013).

Interestingly, already today leading data scient-ists are telling us that Big Data can and must bereduced intelligently to small data, so that finallyfor most decision problems one does not needBig Data at all.4,5

1.5 Collecting the Right (Amount) ofData

Large, global companies already recognise thatthere is a need to stop collecting more data andstart a focused collection of the right data re-quired to make decisions and to run a businesssuccessfully (Nokia Siemens Networks 2013).

Suppose a company is gathering the right data:attributes and dimensions really relevant for plan-ning and decision-making. There is still the ques-tion whether the return on adding more data

4Big Data: Maybe You Don’t Need It : http://www.datacenterjournal.com/it/big-data-dont/

5Most data isn’t big, and businesses are wasting moneypretending it is:http://qz.com/81661/most-data-isnt-big-and-businesses-are-wasting-money-pretending-it-is/

http://www.datacenterjournal.com/it/big-data-dont/

http://www.datacenterjournal.com/it/big-data-dont/

http://qz.com/81661/most-data-isnt-big-and-businesses-are-wasting-money-pretending-it-is/




points diminishes after passing a certain volumeof data collection, or certain data granularities(such as monitoring intervals), and if – in a par-ticular situation – gathering additional data willcost more than it will actually yield.

Cleary, an answer to that question depends onthe concrete enterprise planning and decisionproblem, the importance of the problem, thescalability of engines/algorithms processing thedata, the tolerance of the algorithms regardingartifacts and noise, the skills of the managersprocessing and interpreting the data, and manymore factors.

However, independent of particular problemsand individual factors as aforementioned, the an-swer also depends on purely statistical or math-ematical criteria regarding redundancy and noisewithin the datasets. That is because such criteriacan determine if another piece of data can bringnovel information at all, or whether it can befully or approximately derived from data alreadyavailable (for example by means of collaborativemechanisms such as regression or causal reason-ing).

Furthermore, for reasons of robustness and scalab-ility it is disadvantageous to parametrise predic-tion models and mathematical decision programswith correlated or even collinear data vectors.In fact, efficient decision mechanisms should berather elastic and adaptable to the anatomy andthe information contained in the input data, whiletoday typically the signatures and internal al-gorithms of enterprise decision modules are ofrather static nature.

Consider a resource allocation mechanism forenterprise services in a data centre. If demandforecasts were expected to be highly precise forcertain indicators over a defined period of time,a rather aggressive allocation mechanism oper-ating with deterministic demand curves wouldbe appropriate. Once the demand prediction tooldowngrades its confidence levels and shrinks thehorizon of the look-ahead period considered as

reliable, more conservative allocation mechan-isms might be appropriate.

If the forecasting horizon approaches zero timeintervals, conservative online mechanism shouldbe applied that allow for handling unexpected de-mand phases immediately, as sophisticated offline-planning would not beneficial in such situations:plans would be invalid shortly after their compu-tation.

This paper reviews theory and practice of datareduction in service management with regard tothe various targets addressed with the differentdata reduction techniques. First, we argue thata really efficient and intelligent data reductionrequires the prior definition of business problemsand algorithms how to address these problemswith reduced data. Second, we argue that math-ematical programs and algorithms for planningand decision-making should not be applied in adata-agnostic fashion. In contrast, programs andalgorithms should be sensitive and adjustable toavailable data and the amount of dependencies,reliability, and stochastics within data, whichtypically vary over time, use-case, domain, andplanning horizon.

2 Data Understanding and Reduction

The first and most important step in analytics isa proper understanding of the available data, theinvolved variables and how these are measured.Data quality, appropriate data cleaning and hand-ling missing values as well as detecting outliersand errors must be performed prior to any dataanalysis. Knowing that data preprocessing is ar-guable the most complex and time-consumingstep in analytics, for now we assume these taskshave been already performed.

We will now characterise various techniques toreduce data to relevant features, structures, anddevelopments. In order to separate approachesaimed at descriptive, predictive, and prescript-ive analytics, we will group the techniques ac-cordingly. Descriptive analytics will be furtherdifferentiated in simple aggregations (Sect. 2.1),


110 Thomas Setzer

and approaches that exploit statistical dependen-cies in and between data objects and variables(Sect. 2.2). In Sect. 2.3, we focus on data miningapproaches aimed at gaining knowledge from thedata to reduce uncertainty regarding the realisa-tion of a particular variable (or label). A typicaltask would be the determination of the probab-ility of a positive response of a customer, andthe determination of data (features) necessaryto learn this probability. In Sect. 2.4 we thensummarise approaches to predict whole vectorsor time series. Finally, in Sect. 2.5, we focus onprescriptive data reduction techniques that dif-fer from prescriptive techniques as data selectionand reduction needs to be aligned with a particu-lar, potentially combinatorial and computationalvery complex mathematical optimisation prob-lem. In the latter case, the goal is not only togain insights and reduce uncertainty of futurevalues of data, but to select and transform datain a way that is beneficial for solving a particularplanning and decision problem

2.1 Data Aggregation for DescriptiveService Analytics

The purpose of aggregating data for descriptiveservice analytics is to summarise what happenedin the past. For example, in Web analytics met-rics are considered such as number of page views,conversion rates, check-ins, churns, etc. Thereare literally thousands of such metrics, on theirown typically simple event counters. Other ag-gregations for descriptive service analytics mightbe the results of simple arithmetic operations,such as share of voice, average throughput, aver-age number of positive responds to a campaign,etc. Most of what the industry called analyticsis nothing but applying filters on the data beforecomputing the descriptive statistics, sometimescombined with a linear statistical forecast. Forexample, by applying a geo-filter first, a companycan get metrics such as average revenue per weekfrom USA vs. average revenue per week fromEurope. Structuring aggregated data to reports

derives the well established and broadly used re-porting functions based on information storedin data warehouses. Management dashboardsusually provide the means of presenting suchaggregated data to managers to support theirbusiness decisions.

2.2 Data Compression andApproximation

The most generic way to reduce (and not justaggregate) data is to exploit dependencies in andbetween data vectors – in a problem-agnosticway – by multivariate statistics and matrix ap-proximation techniques, mostly based on linearalgebra. Examples are variance-preserving ap-proximation techniques such as Empirical Ortho-gonal Defactorisation derived by Eigen-approachessuch as Truncated Singular Value Decompositionor compact Principal Component Analysis (PCA).More and more, techniques such as Independ-ent Component Analysis (ICA) are applied to de-rive more meaningful features (in contrast tosolely reducing data). By exploiting communal-ities, such techniques are very useful to reducedata to the maximum amount of variation (as aproxy for information) in the data sets and areoften shown to derive the best low-dimensionalapproximation of data in very useful mathemat-ical senses such as the L2 norm.

Other examples are topology-preserving tech-niques such as Local-Linear-Embedding (LLE) (Ro-weis and Saul 2000) or isoMap (Tenenbaum et al.2000), where the objective of data reduction isnot to capture maximum variance of the datasets with fewer dimensions, but to preserve thetopology of the data objects, i.e., their distancerelationships.

Likewise, multivariate techniques such as vectorquantisation and linear and non-linear regres-sion techniques fall into this category of datareduction according to pre-defined mathematicalobjectives.


2.3 Data Reduction by InformationGain and other Criteria

Unlike the approaches described in Sect. 2.1 andSect. 2.2, the analysis step of discovering know-ledge in databases is aimed at discovering pat-terns in sets of data involving methods at the in-tersection of artificial intelligence, machine learn-ing, statistics, and database systems. The over-all goal is to extract pattern in a data set andtransform it into structural dependencies for fur-ther use. Aside from the raw analysis step, itinvolves database and data management aspects,inference considerations, interestingness metrics,complexity considerations, post-processing ofidentified structures, visualisation, and on-lineupdating mechanisms.

Typical goals are the automatic or semi-automaticanalyses of large quantities of data to extract pre-viously unknown patterns such as groups of datarecords (segmentation analysis), unusual records(anomaly detection) and dependencies via associ-ation rules, decision trees, or other methods. Forinstance, data mining techniques might identifymultiple groups in the data, which can then beused to obtain more accurate prediction resultsand more focused marketing campaigns by a de-cision support system. Here, data is reduced togain information about the general structure ofthe data (clustering), or the class prediction ofrecords with an unknown label due to similarit-ies with other records where labels are alreadyknown.

As discussed in Sect. 1.4, clustering and classifica-tion do not perform well with high-dimensionaldata because of the curse of dimensionality. Beyeret al. (1999) and Aggarwal et al. (2001), amongstothers, have shown that standard measures forproximity or distance that are used for stand-ard k-means clustering, are becoming more andmore meaningless with growing dimensionality.To circumvent this problem, approaches as pro-posed in Aggarwal et al. (2001) introduce noveldistance calculations that are still meaningfulleven in high-dimensional data space of 15 dimen-sions and more. Alternative streams of research

(see Tsymbal et al. 2002 as an example) proposeapproaches that do not work (cluster) on originaldata but on reduced data as a result of compres-sion steps as described in Sect. 2.2.

2.4 Data Reduction for PredictiveService Analytics

Predictive analytics is based on information ex-tracted by the three previous data understandingand reduction steps; it uses all of the gained in-sights to make robust prediction of developmentsof important indicators, metrics, and variables(Stewart et al. 2012).

An intuitive way to understand predictive ana-lytics is to apply it to the time domain. The mostfamiliar predictive analytic tool is a time seriesmodel (or any temporal model) that summarisespast trajectories found in the data, and use eitherauto- or (lagged) cross-correlations and regres-sion to extrapolate time series to a future timewhere data is not yet existing. This extrapolationin the time domain is what scientists refer to asforecasting or prediction.

Although predicting the future is a common usecase of predictive analytics, predictive modelsare not limited to predictions in temporal dimen-sions. Such models can theoretically predict any-thing and, hence, predictive analytics are some-what overlapping with data mining and know-ledge extraction as described in Sect. 2.3. Thepredictive power of a model needs to be prop-erly validated by criteria addressing the robust-ness of the prediction such as using pre-whitenedpredictors, perpendicularity of predictors, by us-ing information criteria such as BIC or AIC, andfinally out-of-sample testing using consecutivesamples. The essence of predictive analytics, ingeneral, is that we use existing data to build amodel. Then we use the model to predict datathat doesn’t (yet) exist.

However, only with concrete use cases in termsof business problems in mind, one can decidewhich pieces of information in the data set areultimately relevant for a company, and which


112 Thomas Setzer

pieces are not. This brings one directly to datareduction for prescriptive analytics that will bedescribed in the next subsection.

2.5 Data Reduction for PrescriptiveService Analytics

Prescriptive analytics not only predicts a possiblefuture, it predicts multiple futures based on thedecision maker’s actions. Therefore a prescript-ive model is, by definition, also predictive andsignificant effort must be undertaken to guaran-tee internal and external model validity. As it isseen today, a prescriptive model is actually a com-bination of multiple predictive models running inparallel, one for each possible input. Since a pre-scriptive model is able to predict the possible con-sequences based on different choices of action,it can also recommend the best course of actionfor any pre-specified outcome, given the dataset used to predict the future (together with itsconfidence or uncertainty). The goal of most pre-scriptive analytics is to guide the decision makertowards decisions that will ultimately lead to an(near) optimal and robust business outcome.

In prescriptive analytics, one also builds a pre-dictive data model. However, the model musthave two more added components in order tobe prescriptive. A company not only needs arigorously validated predictive model, the modelmust be actionable, i.e., managers must be ableto take actions supported by the model. In addi-tion, the prescriptive model must have a feedbacksystem that collects feedback data for each typeof action, which will additionally increase datavolume by some orders of magnitude. There-fore, prescriptive analytics is very challengingeven with scalable data infrastructures and thetalent/expertise to make sense of the feedbackdata (e.g., sensitivity analysis, causal inference,or risk models).

That makes prior data reduction even more im-portant and requires a focus on the pieces ofinput data really relevant for decision-makingand optimisation.

3 Information Gain versusOptimisation Gain

Each department of a service provider has a setof typical tasks to perform on an operational,tactical, or strategic level. Taking for instancethe Customer Relationship Management (CRM)department. CRM is aimed at the optimisation ofa company’s interactions with current and futurecustomers. Objectives of CRM are the reductionof overall churn by adequate customer serviceand support, or by identifying and rewardingcustomers that have been loyal over a period oftime but now show certain behaviours that in-crease churn probability (reduced call frequency,churns of neighbor nodes in the telecommunic-ation network, etc.) Another objective mightbe the identification of customer segments forparticular campaigns such as cross-selling offersbased on score-values of customers. Scores arederived by data analytics and reflect the probab-ility of a certain customer to respond positivelydepending on a customer’s profile and past be-haviour. Such procedures are aimed at gaininginformation from datasets regarding the prob-ability of an unknown label in data records (forinstance, class predictions such as churn: yes/no,upselling: yes/no, etc.) and are in the primaryfocus of business intelligence solutions.

However, usually strict business rules exist thatcomplicate the selection of target customers. Asa simple example, consider the case where onesingle customer is not allowed to be contactedmore than twice a year (a common rule-type intelecommunications companies’ campaign man-agement). This in fact leads to predictive andfinally to prescriptive analytics, as combinatorialdecision problems based on expected behaviouraldevelopments of customers are required (bey-ond the calculation of current scores). Besides acustomer’s score-value for a planned campaign,knowledge of future campaigns are of import-ance as well as on future developments of cus-tomers in order to predict their responses. Inaddition, it has been shown by Goel and Gold-stein (2013), amongst others, that the structure


of the communication or social network and theprediction of future behaviour of a customer’sneighbors play important roles, which brings adecision maker to network models, multivariateforecasting models and collaborative prediction.

While there is a huge body of knowledge ofbroadly used methods and sophisticated toolsexist to perform individual tasks such as classific-ation, time series prediction, or mathematical op-timisation, the integration of these tasks to deriveefficient and robust overall solutions is still leftto the expertise and preferences of individual de-cision makers, typically based on trial-and-errorprocedures or rules of thumb.

For each task, different data reduction techniquesand feature-combination might be adequate,while the interplay of these tasks might lead tothe fact that certain data considered as highlyrelevant in one task might not or only slightlyimpact the overall solution (and vice versa). Forinstance, it might turn out that the prediction offeatures relevant to compute current scores aretoo difficult to predict for future campaigns andthe forecast cannot be considered as reliable. For-mulating a stochastic optimisation model mightreveal that the solution is highly sensitive to evensmall planning errors or rather insensible to lar-ger ones, which makes the predictability of afeature either less or more important. Hence,each type of problem requires individual dataand model selection procedures if the goal is tomake optimal decisions.

This leads to a novel concept in prescriptive ana-lytics that we will refer to as optimisation gainof data. Optimisation gain differs from inform-ation gain (or derivatives such as informationgain rations, GINI, etc.) or matrix approxima-tion quality norms of a residual matrix. Thosemetrics are aimed at quantifying the quality ofa data prediction or approximation without con-textual knowledge on how information is used insubsequent optimisation steps.

By optimisation gain we mean the dependencyof a solution (the solution quality) derived by a

mathematical model or algorithm to additionaldata, which might be more fine-grained data,more data in terms of a longer reliable planninghorizon, or simply an additional attribute or di-mension under consideration.

Optimisation gain also differs from concepts suchas sensitivity, robustness, or stability of a solu-tion. With optimisation gain we address the dif-ferent and more general problem of quantifying,if (and how much) the optimality or robustnessof a solution would benefit for example from theconsideration of a novel data feature in a partic-ular planning or decision problem. Addressingsuch questions is challenging as this typicallyrequires the re-formulation of the mathematicalprogram formulation for numerous input-datacombinations and transformations. The intuitionof optimisation gain is the quantification of thesolution quality expected with different inputdata for a particular type of optimisation prob-lem analytically, without expensive and time-consuming (and potentially infeasible) trial-and-error-procedures. The vision is a new generationof criteria by integrating data and model selec-tion and configuration.

Please notice that optimisation gain can becomenegative as too many parameters can lead to anexplosion of the search spaces and increased com-plexity, where optimal solutions are much harderto find. For instance, node-sets of branch & cutsolvers might increase dramatically, and the qual-ity of solutions that can be found in pre-definedperiods of time might decline sharply with thenumber of features and constraints under consid-eration. Furthermore, models operating with toomany data dimensions are more likely subjectto over-fitting as artefacts and collinear config-urations of (stochastic) variables used as model-input worsen the quality of decision-making.From a business perspective, the marginal gain ofconsidering more data might further decline ascollecting and managing data comes at additionalcosts for data scientists that need to analyse thedata, as well as costs for monitoring, IT infra-structures, storage, and licenses.


114 Thomas Setzer

We argue that the role of optimisation gain ofdata is a highly relevant concept in prescript-ive analytics, and key to reducing Big Data ef-ficiently to a manageable and actionable set offeatures Also, INFORMS, the leading scientificand professional organisation for OR profession-als, decided to stake its claim on the analyticsmovement. The organisation recognised that thetrend toward data-driven and analytical decision-making presents tremendous opportunities andchallenges for OR professionals (Libertore andLuo 2011). Since 2009, INFORMS organises anown conference at the intersection of analyticsand OR named Business Analytics and OperationsResearch, with a focus on how to apply data sci-ence to ‘the art of’ business optimisation. It fea-tures presentations on real-world applicationsof analytic solutions, presented by industry anduniversity leaders.

Optimisation gain can provide a means of signi-ficantly reduce the effort spent for monitoring,collecting and managing data, as ideally onlydata is collected that is indeed supposed to im-prove decisions. Unnecessary frequent measure-ments are also avoided as the collection of correl-ated data that is (statistically) already capturedby other variables. These ideas are closely re-lated to visions such as smart measurement andcollaborative monitoring systems, but with anadditional focus on the impact on the businessrelevance of gathered data. We will further detailon this in Sect. 4.

4 Feature-based Optimisation andModel-Data-Integration

As aforementioned, certain units in enterpriseshave specific tasks to perform, usually composedby structured or at least semi-structures pro-cesses. For instance, in IT service management,the role of capacity management is to ensuresufficient capacity to provide high-quality ser-vices to customers efficiently, i.e., at reasonable(low) costs to the business. In capacity manage-ment, it is important to have a clear picture of

the expected service demand and the correspond-ing resource demand that needs to be suppliedin future points of time. Considering the caseof private clouds, with the potential of hostingservices in virtual machines (VM) in a flexiblemanner, e.g., by co-hosting VMs temporarily onthe same physical server, sharing and multiplex-ing a servers capacity for resources such as CPU,memory, or I/O. In such an environment, IT ser-vice managers try to minimise the number ofservers by assigning enterprise services in vir-tual machines efficiently to physical servers, butat the same time provide sufficient computing re-sources at each point in time. It is worthwhile tonotice that running servers are (independent oftheir utilisation levels) the main energy driversin data centers, where energy costs already ac-count for 50% or even more of total operationalcosts (Filani et al. 2008).

Without going into too much detail, the result-ing VM allocation problem can be reduced to astochastic multi-dimensional bin-packing prob-lem, a well-known NP-hard problem. As it is thecase with every bin-packing problem, the goalis to fill-up the available spaces (resource capa-cities) of bins (servers) as much as possible, and,hence, come out with fewer servers while notexceeding the capacity of servers, as this wouldresult in overload and SLA violations.

Theoretically, historical workload data would al-low for accurate workload demand forecasting(for more than 80% of typical operational busi-ness services) and optimal allocation of enter-prise applications to servers. In various exper-iments and studies with smaller VM sets it hasbeen shown that such approaches lead to a reduc-tion of required server by around 30% (Speitkampand Bichler 2010). Unfortunately the volume ofdata and the large number of resulting capacityconstraints in a mathematical problem formu-lation renders this task impossible for any butsmall instances and is of little use for IT serviceproviders with server parks of hundreds or thou-sands of VMs to be consolidated.


Looking at the core of each packing problem, inparticular at bin-packing problems, the challengeis to find complementarity in objects to be packed(in our case, the demand profiles of VMs for vari-ous resources over time) to achieve high averageserver utilisation levels. It makes sense to co-hostVMs with peak loads in the morning hours andVM having their peak loads later during a day.Similarly it makes sense to combine a VM withhigh CPU and low memory demand with onehaving lower CPU but high memory demand.

When we consider relevant features of workloadprofiles for the packing problem as aforemen-tioned, features describing the complementarit-ies between VM profiles could be of great value,besides features describing the absolute resourcedemand curves of VMs.

Setzer and Bichler (2012) use techniques based onsingular value decomposition (SVD) to extractsignificant features from a matrix of the expec-ted (fine-grained) demand vectors of hundredsof VMs and provide a new geometric interpret-ation of these features as principal demand pat-terns, complementary between these patterns,and uncertainty. The extracted features allowfor formulating a much smaller allocation modelbased on integer programming and allocatinglarge sets of applications efficiently to physicalservers. While SVD is typically applied for ana-lytical purposes only such as time series decom-position, noise filtering, or clustering, here fea-tures are used to transform a high-dimensionalallocation problem in a low-dimensional integerprogram with only the extracted features in amuch smaller constraint matrix. The approachhas been evaluated using workload data from alarge IT service provider and results show that itleads to high solution quality. At the same timeit allows for solving considerably larger probleminstances than what would be possible withoutprescriptive analytics, intelligent data reductionand model transform. This work provides a firstexample of a highly integrated data reductionand optimisation approach.

The same authors argue that the overall approachcan also be applied to other large packing prob-lems. For instance, in Setzer (2013), the authorsshow that high-dimensional knapsack problemscan also be intelligently reduced to smaller andcomputationally tractable ones, as long as there isa significant amount of shared variance amongstthe dimensions to be considered. Please noticethat, according to recent studies, knapsack-prob-lems are amongst the top four problems to besolved in enterprises, although managers oftendo not know that their particular problems couldbe formulated as knapsack-problems.

Overall, we believe that there is a huge poten-tial for solving particular decision problem withBig Data made small. However, to exploit thesepotentials, problems must be formalised beforeintegrated data reduction and optimisation mod-els can be developed.

Reconsidering the example of capacity manage-ment in private cloud infrastructures, we willnow detail on the need for a decision model fab-ric that not only aligns the model to be usedto changing environments by considering novelparameters. In contrast, completely differentsolution techniques are required depending onthe (recent) structures and developments foundin the data. Again, we will use private clouds forillustration.

Nowadays, live migration allows to move VMsto other servers reliably even during runtimeand promises further efficiency gains (VMWareESX, amongst others) (Nelson et al. 2005). Someplatforms such as VMware or vSphere closelymonitor the server infrastructure in order to de-tect resource bottlenecks by tracking threshold-violations. If such a bottleneck is detected theytake actions to dissolve it by migrating VMs todifferent servers. For instance, if the CPU utilisa-tion exceeds 80%, a VM is migrated away fromthat server to reduce total server load. On theother hand, if a controller detects phases of lowoverall workload, there is the possibility to con-centrate workloads on fewer servers by vacating


116 Thomas Setzer

servers and shutting down these source serverstemporarily to further reduce energy consump-tion. We will refer to such techniques as dynamicresource allocation or dynamic control, as op-posed to static VM allocation where allocationsare computed and kept fixed for a longer periodof time.

5 Towards Data-ElasticDecision-Making

On the one hand, dynamic control strategiesare more flexible and should therefore lead tolower energy costs. On the other hand, migra-tions cause significant additional overheads andresponse-time peak, which are avoided with staticallocation mechanisms. It has been shown thatwith well-predictable workloads of business ap-plications, dynamic resource allocation duringoperational business hours does not lead to higherenergy efficiency compared to static allocationeven if future demand is known only to a certainextend (Wolke et al. 2013). However, if demandis completely unknown, dynamic control is theonly reasonable option to avoid both: massiveoverprovisioning and service degradation. De-pending on the share of stochastic developmentsin workload demand curves, hybrid models mightbe appropriate where basic allocations are com-puted for a given planning horizon in a moreconservative fashion, considering the option ofpotential migrations to cope with uncertainty.

In summary, dynamic, data-based model selec-tion is required that differs from parameter align-ment, which simply would mean that for instancethe alpha parameter in an exponential smoothingmodel is adjusted from time to time (which thenleads to a different and hopefully better shortterm prediction), but where the same mathemat-ical model is used for prediction.

In the example above, depending on the predict-ability of demand behaviour, which might bewell predictable throughout certain periods butrather unpredictable in other periods of time,completely different allocation mechanisms areadvised.

6 Conclusion and Vision

Analysing historical and current data in order tomake better predictions is vital for running a com-petitive service company. Data-driven designand management of services demand interdis-ciplinary knowledge from the business domain,processes, data analytics, and mathematical op-timisation. While in principle the ever-growingamounts of available data would allow for de-riving increasingly precise forecasts and optim-ised input for planning and decision models, thecomplexity resulting from the consideration oflarge volumes of ever-growing volumes of mul-tivariate, fine-grained data leads to the fact thatdependencies and relationships within the dataare not found, algorithms do not scale, and tra-ditional statistics as well as data-mining tech-niques collapse because of the well-known curseof dimensionality. Hence, in order to make BigData actionable, we are interested in the intelli-gent reduction of vast amounts of data to smallsets of problem-relevant features. We argue thatmathematical optimisation and planning mod-els need to be transformed to be able to operateefficiently on highly reduced data. In addition,the selection of adequate planning and decisionmodels must be adapted to (current) data andthe reliability of relations and predictions extrac-ted from that data, which requires time-dynamicand data-driven model selection and evaluationtechniques.

References

Accenture (2006) Service Management – En-abling High Performance Through SupplyChain Management. Accenture.

Aggarwal C. C., Hinneburg A., Keim D. A. (2001)On the Surprising Behavior of Distance Met-rics in High Dimensional Space. In: DatabaseTheory – ICDT 2001. LNCS. Springer, London,pp. 420–434

Bellman R. E. (1961) Adaptive Control Processes:A Guided Tour. Princeton University Press


Beyer K. S., Goldstein J., Ramakrishnan R., ShaftU. (1999) When is ’Nearest Neighbor’ Mean-ingful. In: Proc. Int. Conf. Database Theory(ICDT ’99). Springer, London, pp. 217–235

Choi D., Ahn B. S. (2009) Eliciting CustomerPreferences for Products From Navigation Be-havior on the Web: A Multicriteria DecisionApproach With Implicit Feedback. In: IEEETransactions on Systems, Man and Cyber-netics, Part A: Systems and Humans 39(4),pp. 744–760

Davenport T. H., Harris J. G. (2006) Competingon Analytics: The New Science of Winning.Harvard Business Review Press

Davenport T. H., Patil D. J. (2012) Data Scientist:The Sexiest Job of the 21st Century. http ://hbr.org/2012/10/data-scientist-the-sexiest-job-of-the-21st-century/

Filani D., He J., Gao S., Rajappa M., Kumar A.,Shah P., Nagappan R. (2008) Dynamic DataCenter Power Management: Trends, Issues,and Solutions. In: Intel Technology Journal

Goel G., Goldstein D. G. (2013) Predicting In-dividual Behavior with Social Networks. In:Marketing Science

Lee J. A., Verleysen M. (2007) Nonlinear Dimen-sionality Reduction. Springer, Heidelberg

Libertore M., Luo W. (2011) INFORMS and theAnalytics Movement: The View of the Mem-bership. In: Journal Interfaces 41(6), pp. 578–589

Nelson M., Lim B.-H., Hutchins G. (2005) FastTransparent Migration for Virtual Machines.In: Proc. USENIX. Berkeley, USA, pp. 25–35

Nokia Siemens Networks (2013) Exit BigData. Enter Right Data. http : / / www .nokiasiemensnetworks.com/news- events/insight-newsletter/articles/exit-big-data-enter-right-data

Office of Government Commerce (OGC) (2009)Official ITIL Website. Last Access: [Online].Available: http://www.best-management-practice.com/IT-Service-Management-ITIL

Ross J. W., Beath C. M., Quaadgras A. (2013)You May Not Need Big Data After All. In:Harvard Business Review 91(12)

Roweis S., Saul L. (2000) Nonlinear Dimension-ality Reduction by Locally Linear Embedding.In: Science 290(5500), pp. 2323–2326

Setzer T. (2013) An Angle-Based Approachto Solving High-Dimensional Knapsack-Problems. KIT

Setzer T., Bichler M. (2012) Using Matrix Ap-proximation for High-Dimensional DiscreteOptimization Problems. In: European Journalof Operational Research 227, pp. 62–75

Speitkamp B., Bichler M. (2010) A MathematicalProgramming Approach for Server Consolid-ation Problems in Virtualized Data Centers.In: IEEE TSC 3(4), pp. 266–278

Stewart T., Thomas R., McMillan C. (2012) De-scriptive and Prescriptive Models for Judg-ment and Decision Making: Implications forKnowledge Engineering. In: Expert Judgmentand Expert Systems – NATO AS1 Senes 35,pp. 314–318

Tenenbaum J. B., Silva V. d., Langford J. C. (2000)A Global Geometric Framework for Nonlin-ear Dimensionality Reduction. In: Science290(5500), pp. 2319–2323

The Economist (Feb. 2010) A Special Reporton Managing Information: Data, Data Every-where. In: The Economist

Tsymbal A., Pechenizkiy M., Baumgarten M.,Patterson D. (2002) Eigenvector-Based Fea-ture Extraction for Classification. In: Proc.Int. Artificial Intelligence Research SocietyConference. Florida, USA

Wolke A., Bichler M., Setzer T. (2013) Energyefficient capacity management in virtualizeddata centers: optimization-based planning vs.real-time control. In: Proc. INFORMS Work-shop on Information Systems Technologies.Milan, Italy

Thomas Setzer

Karlsruhe Institute of TechnologyEnglerstraße 1476131 [email protected]

http://hbr.org/2012/10/data-scientist-the-sexiest-job-of-the-21st-century/



http://www.nokiasiemensnetworks.com/news-events/insight-newsletter/articles/exit-big-data-enter-right-data





118 Imprint

Enterprise Modelling and Information Systems Architectures

The journal Enterprise Modelling and Information Systems Architectures is the official journal of the SpecialInterest Group on Modelling Business Information Systems within the German Informatics Society (GI-SIGMoBIS).

The journal Enterprise Modelling and Information Systems Architectures is intended to provide a forumfor those who prefer a design-oriented approach. As the official journal of the German InformaticsSociety (GI-SIG-MoBIS), it is dedicated to promote the study and application of languages and methodsfor enterprise modelling – bridging the gap between theoretical foundations and real world requirements.The journal is not only aimed at researchers and students in Information Systems and Computer Science,but also at information systems professionals in industry, commerce and public administration who areinterested in innovative and inspiring concepts.

The journal’s editorial board consists of scholars and practitioners who are renowned experts on variousaspects of developing, analysing and deploying enterprise models. Besides Information Systems, theycover various fields of Computer Science.

Subscription InformationThe journal is distributed free of charge for members of the GI-SIG-MoBIS. Membership can be acquiredthrough the German Informatics Society (http://www.gi-ev.de/verein/mitgliedschaft/). Single issues,priced at EUR 25 each (plus shipment), can be ordered online (http://www.fg-mobis.gi-ev.de/).

Enterprise Modelling and Information Systems ArchitecturesVol. 9, No. 1, June 2014Editorial Board 119

Editorial Board

Editors in ChiefManfred Reichert, Ulm UniversityKlaus Turowski, Otto von Guericke University Magdeburg

Associate EditorsWil van der Aalst, Eindhoven University of TechnologyWitold Abramowicz, Poznan University of EconomicsColin Atkinson, University of MannheimJörg Becker, University of MünsterJörg Desel, University of HagenWerner Esswein, Dresden University of TechnologyFernand Feltz, Centre de Recherche Public Gabriel LippmannUlrich Frank, University of Duisburg-EssenAndreas Gadatsch, Bonn-Rhine-Sieg University of Applied SciencesMartin Glinz, University of ZurichNorbert Gronau, University of PotsdamWilhelm Hasselbring, University of KielBrian Henderson-Sellers, University of Technology, SydneyStefan Jablonski, University of BayreuthManfred Jeusfeld, Tilburg UniversityReinhard Jung, University of St. GallenDimitris Karagiannis, University of ViennaJohn Krogstie, University of TrondheimThomas Kühne, Victoria University of WellingtonFrank Leymann, University of StuttgartStephen W. Liddle, Brigham Young UniversityPeter Loos, Johannes Gutenberg-University of MainzOscar Pastor López, Universidad Politècnica de ValènciaHeinrich C. Mayr, University of KlagenfurtJan Mendling, Vienna University of Economics and BusinessMarkus Nüttgens, University of HamburgAndreas Oberweis, University of KarlsruheErich Ortner, Darmstadt University of TechnologyErik Proper, Radboud University NijmegenMichael Rebstock, University of Applied Sciences DarmstadtStefanie Rinderle-Ma, University of ViennaMichael Rosemann, Queensland University of TechnologyMatti Rossi, Aalto UniversityElmar J. Sinz, University of BambergFriedrich Steimann, University of HagenStefan Strecker, University of HagenBernhard Thalheim, University of KielOliver Thomas, University of OsnabrückJuha-Pekka Tolvanen, University of JyväskyläGottfried Vossen, University of MünsterMathias Weske, University of PotsdamRobert Winter, University of St. GallenHeinz Züllighoven, University of Hamburg

ISSN 1866-3621 (Online)


120 Guidelines for Authors

Guidelines for Authors

The journal serves to publish results of innovative research on all facets of creating and analysingenterprise models and information systems architectures. For research papers, it is required to satisfyacademic standards in terms of originality, level of abstraction and justification of results. Experiencereports serve to describe and analyse success stories as well as practical obstacles and resulting researchchallenges. Topics covered by the journal include, but are not restricted to the following subjects:

• Languages and Methods for Enterprise Modelling

• Reusable Domain Models (Reference Models)

• Analysis and Design Patterns

• Modelling of Business Processes and Workflows

• Process-Oriented System Architectures

• Component-Oriented System Architectures

• Conceptual Modelling for Component-Oriented Design

• Ontologies for Enterprise Modelling

• Modelling for Enterprise Application Integration

• Modelling for Data Warehouses

• Modelling to support Knowledge Management

• Model-Driven Development

• Aspect-Oriented Design

• Agile Methods for Enterprise Modelling

Authors are asked for electronic submissions, which have to be sent to the editor in chief as e-mailattachment. In case of multiple authors, it is required to name one author who acts as contact person.The submission should include a cover page with the paper’s title and the names, affiliations and e-mailaddresses of all authors. The first page of the paper starts with the title and does not carry the authors’names. A manuscript must be in PDF format. It should not exceed 5.000 words – this includes an abstractof around 150 words.

Submitted papers will be reviewed within no more than two months. The review process is double blind.Authors who submit a manuscript guarantee that it has not been published elsewhere, nor is intendedto be published elsewhere. Papers that were accepted for publication must be written according to thestyle defined for the journal. A comprehensive description as well as a corresponding LaTeX template isprovided on the web portal of the GI-SIG-MobIS (http://www.fg-mobis.gi-ev.de/).

ISSN 1866-3621 (Online)

Documents

Enterprise Modelling and Information Systems Architectures