© Copyright 2015 Good e-Learning. All rights reserved. No part of this publication may be reproduced, resold, stored in a retrieval system, or distributed in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without the prior permission of the copyright owner. Such requests for permission or any other comments relating to the material contained in this document may be submitted to: marketing@goodelearning.com. Good e-Learning is a trading name used by

Educational Systems Ltd.

What Makes BPMN Complex?by Gregor Polancic

IntroductionWhile the objectives of BPMN are to be expressive, widely understood, interchangeable and executable, it has also became rich and complex. Besides, to stay competitive to related standards such as Business Process Execution Language (BPEL), an execution semantic and the corresponding elements were introduced into the BPMN 2.0. BPMN was arguably already a complex notation prior to the addition of these new elements, thus many studies were done with the aim of decreasing its complexity. In this paper, we will firstly define complexity from a general perspective and from a perspective of a notation and a model. Secondly, we will define what makes BPMN complex and finally, we will analyse the previous studies addressing the BPMN complexity.

Notation vs. Model ComplexityWhen addressing complexity of a modelling notation, there are two aspects: the complexity of a notation and the complexity of a model, conducted in accordance with the said notation.

Among other indicators, the complexity of notation can be defined also by the number of unique elements and the rules of their usage. Currently, BPMN 2.0 includes 100+ unique elements and an advanced set of rules regarding their usage.

On the other hand, even if we use just a small scope of available elements, we can still produce a complex model. This is evident in Figure 1, which represents an official example provided by the OMG of a small, yet fairly complex model. Syntactically it incorporates approximately 20 elements. Semantically, it describes e-mail voting process, used for resolving issues through e-mail votes.

BPMN Series #15 | ATL001:15

Figure 1: A complex BPMN model (OMG)

“Complexity of a system is the degree to which a system or component has a design or implementation that is difficult to understand and verify.”

Institute of Electrical and Electronics Engineers - IEEE

“Complexity is the degree to which a system or component has a design or implementation

that is difficult to understand and verify.” Software Engineering Institute - SEI

© Copyright 2015 Good e-Learning. All rights reserved. No part of this publication may be reproduced, resold, stored in a retrieval system, or distributed in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without the prior permission of the copyright owner. Such requests for permission or any other comments relating to the material contained in this document may be submitted to: marketing@goodelearning.com. Good e-Learning is a trading name used by

Educational Systems Ltd.

We can conclude that complexity can be considered subjective and it applies to both the notation and the model. This is important when defining complexity of a notation, since the number of elements and relations alone may not be enough to characterize complexity. A large number of elements and relations can still present a simple system. On the other hand, a few elements and relations can still produce a complex model (Padilla et al. n.d.).

What Makes BPMN Complex?When we are addressing complexity, we have to take into the consideration that it is hard to measure and is considered to be subjective. So, instead of measuring complexity, it is more convenient to look for the signs of complexity. This a method that is borrowed from medicine where signs help to confirm the diagnosis. Accordingly to Saltzer & Kaashoek, there are five sings of complexity, which are summarized in Table 1, along with the examples from BPMN and its corresponding business process diagram (BPD).

As can be seen from Table 1, BPMN is showing all five signs of a complex system and can be thus pronounced as complex.

What do Others Say?The complexity of BPMN is a widely discussed topic also among the researchers. Many studies were done with the goal of analysing how complex BPMN notation is in both theory and practice. Many concluded that BPMN is still perceived as complex and not easy to learn, even for business analysts.

More specifically, Recker performed a world-wide survey in 2007, regarding the usage of BPMN 1.0 in practice, where 590 users responded. The results showed that 36% of respondents used only the core BPMN set, 37% used the extended set of BPMN and only 27% used all of what BPMN has to offer. As to which symbols are used in practice, the author additionally interviewed 20 BPMN users. The results are summarized in Figure 2.

As can be seen from the figure above, Normal Flows, Links and Text Annotations, were rated as essential for process modelling. Contrarily, the most unused BPMN elements were Off-page connector, Group, and Multiple. They were classified by over 50% as being “not in use”, “not understood” or “not aware”.

Based on the findings above, BPMN was recognized as being rich yet complex and not easy to learn. Additionally, some concerns were raised by the author about the necessity to extend the BPMN element set in BPMN 1.2, especially Events. Author questioned: “Really, is this necessary? Isn’t there enough already?”

Table 1: Five signs of complexity

Sign of Complexity BPMN / BPD

Large number of components - a size has a direct influence of whether we perceive a system complex or not.

BPMN 2.0 has over 110 elements.

Large number of interconnections - even when we operate with a small amount of components, we can have many interconnections between them, making a system arguably complex.

In a BPD, BPMN elements need to be interrelated into a graph.

Many irregularities - we can have a large number of components and interconnections, yet they can still form a simple system. However, if there are many exceptions or irregularities in the system, we can perceive this as a sign of complexity.

Vertical BPMN domains, the meaning of non-standardized elements, the meaning of colours, etc.

A long description - to avoid complexity a methodological description of the system is advised. A shorter, systematic specification is favoured over a long description of the system’s properties.

BPMN 2.0 specification is 538 pages long.

A team of designers, implementers or maintainers of a system. Usually there are several roles involved in constructing and maintaining a system, so it should be simple enough for each individual to understand it.

Business process modelling involves different business and technical roles.

Figure 2: Use of BPMN elements (Recker, 2007)

© Copyright 2015 Good e-Learning. All rights reserved. No part of this publication may be reproduced, resold, stored in a retrieval system, or distributed in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without the prior permission of the copyright owner. Such requests for permission or any other comments relating to the material contained in this document may be submitted to: marketing@goodelearning.com. Good e-Learning is a trading name used by

Educational Systems Ltd.

This was later confirmed in 2010 by Leymann, where author argues that the complexity has further increased in BPMN 2.0. This is partially due to the support of execution of BPMN 2.0 models and the addition of features, required by the domain users. Figure 3 illustrates how the goals of different versions of BPMN are related to its complexity.

BPMN was again identified as being rich in a case study of its use in practice, done by Muehlen and Ho in 2008. The focal point of the case study was to analyse the service management process in a truck dealership. The as-is and to-be processes were documented in BPMN, whereas the financial impact of the proposed changes was done using a BPMN-based simulation tool. Even though the study reports that the participants were being able to read the diagrams with relative ease, it should be noted that the layout of the process was designed as simple as possible and a limited set of BPMN constructs were used, as is represented in Figure 4.

Also, despite the introduction of the entire set of BPMN elements, the audience remembered only the basic elements.

That users usually limit themselves to only basic constructs was once again confirmed in a study by Meuhlen and Recker, where the authors analysed 120 BPMN models. The study has shown that the average model consisted of only 9 different elements, which translates to roughly 20% of all BPMN 1.x elements. Additionally, BPMN was compared to Unified Modeling Language (UML), where the complexity of the language has forced modellers to reduce the set of elements they use to a very basic group. Similar was concluded for BPMN, too. Authors have concluded that complexity of BPMN in practice differs from its theoretical complexity.

Based on the previous researches that have been done in the field of BPMN usage, we can conclude that BPMN is a rich yet complex notation. The theoretical complexity of BPMN is high, especially with the introduction of new diagraming techniques and new elements. However, in practice, only a small subset of elements are commonly used, suggesting that the complexity in practice is not as severe as the complexity in theory.

Figure 3: Complexity conflict between different BPMN objectives

Cus

tom

erTr

uck

man

ufac

ture

r

Serv

ice

Advis

orFo

rem

anTe

chni

cian

Truck needs service

Call service shop

Drive truck to the site

Acknowledge estimated

cost & term

t(0) Answer question for customer

t(1) Open R.O.

t(2) Analyse complaint

t(3) Write jobs

t(4) Send R.O. to foreman

t(5) Move truck to

repair site

g(0) Specific service

requested?

t(6) Assign diagnosis job to

technician

t(7) Perform/confirm diagnosis

t(8) Report back to S.A.

t(9) Re-analyse the

caseYes

g(1) Different problem domain?

No

Yes

No

Estimated problem/cost/wait time

Request for service

Conversation

R.O.

Customer complaint

Figure 4: An excerpt from the as-is model

© Copyright 2015 Good e-Learning. All rights reserved. No part of this publication may be reproduced, resold, stored in a retrieval system, or distributed in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without the prior permission of the copyright owner. Such requests for permission or any other comments relating to the material contained in this document may be submitted to: marketing@goodelearning.com. Good e-Learning is a trading name used by

Educational Systems Ltd.

ConclusionBPMN is the de facto standard for business process modelling, which enables users to model from completely basic to advanced business process models. As such, it provides a rich notation with a well over 100 symbols, covering from basic modelling to advanced execution requirements. Consequentially, it shows the fives signs of complexity and has been reported as such by many business users. Much of research was done in light of exploring the theoretical and practical complexity of the notation and the results consistently indicate that BPMN is indeed a complex notation, especially not understood well by beginners. However, it has been shown that in practice, not many elements are used, displaying that the theoretical complexity is much higher than the practical one. Still, in light of successfully addressing the complexity, Saltzer and Kaashoek proposed four complexity management techniques, namely modularity, abstraction, layering and hierarchy, which will be discussed in following papers.

ReferencesLeymann, F., 2010. BPEL vs. BPMN 2.0: Should You Care? In Jan Mendling, M. Weidlich, & M. Weske, eds. Business Process Modeling Notation. Berlin, Heidelberg: Springer Berlin Heidelberg, pp. 8-13. http://www.springerlink.com/index/10.1007/978-3-642-16298-5_2

Muehlen, M. zur & Ho, D.T., 2008. Service Process Innovation: A Case Study of BPMN in Practice. In Proceedings of the 41st Annual Hawaii International Conference on System Sciences (HICSS 2008). 41st Annual Hawaii International Conference on System Sciences (HICSS 2008). Waikoloa, HI, USA, pp. 372-372. http://ieeexplore.ieee.org/lpdocs/epic03/wrapper.htm?arnumber=4439077

Muehlen, M. zur & Recker, J.C., 2008. How Much Language is Enough? Theoretical and Practical Use of the Business Process Modeling Notation. In Berlin :: Springer,. http://www.bpm-research.com/download/papers/MIZU-JARE-BPMN-CAiSE-2008.pdf

Padilla, J.J. et al., Towards a Diagnostic Framework for Understanding Complex Situations, Norfolk, VA: Old Dominion University. http://www.necsi.edu/events/iccs7/papers/163PadillaOther.pdf

Recker, J.C., 2008. BPMN Modeling – Who, Where, How and Why. http://www.sparxsystems.com.au/press/articles/pdf/bpmn_survey.pdf

Saltzer, J. & Kaashoek, F., 2009. Principles of computer system design: An introduction, Burlington, MA :: Morgan Kaufmann.