mics

h t

Ch

ent,

03; ac

a c

y pre

the

attr

facilitating a group decision process. A real-world example demonstrates the feasibility of the proposed framework.

2004, Yao and He, 2000). A successful ERPproject involves managing business process

tionalities or all special business requirements(Sarkis and Sundarraj, 2000; Teltumbde, 2000;

ARTICLE IN PRESSchange, selecting an ERP software system and a Hong and Kim, 2002). Therefore, companies mustchoose a exible ERP system and a co-operativevendor that is responsive to customer needs. Alltoo often there is no systematic evaluation frame-work in place when most companies evaluate ERP

*Corresponding author. Tel.: +886-3-5742655; fax: +886-3-

5722685.

E-mail address: mjwang@ie.nthu.edu.tw (M.-J.J. Wang).

0925-5273/$ - see front matter r 2004 Elsevier B.V. All rights reserved.doi:10.1016/j.ijpeKeywords: ERP system; Decision analysis; Analytic hierarchy process (AHP); Information system

1. Introduction

Severe market competition has dramaticallytransformed the business environment with theresult that companies need to reduce total costs,maximize return on investment, shorten leadtimes, and be more responsive to customerdemands. Highly dynamic markets call for effec-tive enterprise information systems to enhancecompetitive advantage. Enterprise Resource Plan-ning (ERP) is increasingly important in modernbusiness because of its ability to integrate the owof material, nance, and information and tosupport organizational strategies (Yusuf et al.,

co-operative vendor, implementing this system,and examining the practicality of the new system.Owing to the complexity of the business environ-ment, the limitations in available resources, andthe diversity of ERP alternatives, ERP systemselection is tedious and time consuming. However,given the considerable nancial investment andpotential risks and benets, the importance of apertinent ERP system selection cannot be over-emphasized (Teltumbde, 2000).Existing ERP commercial packages cannot

provide a once-for-all business model for everyprocess of all industry. Thus, no single ERPpackaged software can meet all company func-r 2004 Elsevier B.V. All rights reserved.Int. J. Production Econo

An AHP-based approac

Chun-Chin Wei, Chen-Fu

Department of Industrial Engineering and Engineering Managem

Received 28 March 20

Abstract

An Enterprise Resource Planning (ERP) system is

competitiveness and performance of a company. This stud

ERP system. The framework can systematically construct

and strategies of an enterprise, identify the appropriate.2004.03.00496 (2005) 4762

o ERP system selection

ien, Mao-Jiun J. Wang*

National Tsing Hua University, Hsin Chu, Taiwan 30043, ROC

cepted 5 March 2004

ritical investment that can signicantly affect future

sents a comprehensive framework for selecting a suitable

objectives of ERP selection to support the business goals

ibutes, and set up a consistent evaluation standard for

ARTICLE IN PRESS

uctionsystems. In addition, ERP vendor hype furthercomplicates the selection process. Decision makersfrequently adopt the common ERP evaluationcriteria as the measures without developing tailor-made objectives and clear requirements that echothe company characteristics, its position in itscompetitive environment, and its corporate strat-egy. The result is an inevitable delay of ERPimplementation and under-performance of thesystem. Hence, an ERP system selection frame-work is extremely critical in assisting executivesto evaluate from the perspective of companystrategies.Since the business environment is characterized

by high uncertainty, the process of ERP systemassessment involves numerous problems. Kumaret al. (2002) emphasized that installing an ERPsystem is much more than having another infor-mation technology tool; it is a decision on how toshape the organizational business. Motwani et al.(2002) emphasized that ERP adoption involvesinitiating appropriate business process changes aswell as information technology changes to sig-nicantly enhance performance, quality, costs,exibility, and responsiveness. However, manycompanies install their ERP systems hurriedlywithout fully understanding the implicationsfor their business or the need for compatibilitywith overall organizational goals and strategies(Hicks and Stecke, 1995). The result of thishasty approach is failed projects or weaksystems whose logic conicts with organizationalgoals.This study proposes a comprehensive ERP

system selection framework in which the objectivehierarchy is constructed and the appropriateattributes are specied to provide detailedguidance for ERP system evaluation. The pro-posed methodology also ensures that theevaluation process is aligned with the competitivestrategies and goals of the enterprise. Theanalytic hierarchy process (AHP) method (Saaty,1980) is applied for dealing with the ambiguitiesinvolved in the assessment of ERP alternatives andrelative importance weightings of attributes. Anempirical case in Taiwan is described to demon-strate the practical viability of the proposed

C.-C. Wei et al. / Int. J. Prod48method.2. Selection method review

A number of methods have been applied to ERPor other information system (IS) selection includ-ing scoring, ranking, mathematical optimization,and multi-criteria decision analysis. The scoring(Lucas and Moore, 1976) method is intuitive, buttoo simple to truly reect opinions of the decisionmakers. Buss (1983) proposed a ranking approachto compare computer projects. This method alsohas the same limitation with scoring method.Mathematical optimization such as goal program-ming, 01 programming, and nonlinear program-ming have been applied to resource optimizationfor IS selection. Santhanam and Kyparisis (1995,1996) proposed a nonlinear programming modelto optimize resource allocation allowing for theinteraction of factors; their model consideredinterdependencies between projects in the ISselection process. Lee and Kim (2000) claimedthat Santhanam and Kyparisis model dealt withIS selection problems with limited criteria. Theycombined the analytic network process (ANP) anda 01 goal-programming model to select an ISproject. Badri et al. (2001) presented a 01 goalprogramming model to select an IS projectconsidering multiple criteria including benets,hardware, software and other costs, risk factors,preferences of decision makers and users, comple-tion time, and training time constraints. However,the applicability of these methods is often wea-kened by sophisticated mathematic models orlimited attributes to carry out in a real-worldERP system selection decision, especially whensome attributes are not readily quantiable, as wellas not too easy for managers to understand.Moreover, these methodologies focus too muchon quantiable calculations and look down uponthe comprehensive selection framework of ERPsystem and the strategic considerations of acompany.The AHP method, introduced by Saaty (1980),

directs how to determine the priority of a set ofalternatives and the relative importance of attri-butes in a multiple criteria decision-making pro-blem, and has been widely discussed in variousaspects. For example, Schniederjans and Wilson

Economics 96 (2005) 4762(1991) utilized the AHP method to determine the

to select

criteria

fulllme

gible measures with respect to numerous competing

ARTICLE IN PRESS

uctionobjectives inherent in ERP system selection andfacilitates the group decision-making process.

3. Procedure for selecting a suitable ERP system

To clearly present the proposed ERP systemselection framework, a stepwise procedure is rstdescribed:

Step 1. Form a project team and collect allpossible information about ERP vendorsand systems.

Step 2. Identify the ERP system characteristics.Step 3. Construct a structure of objectives to

develop the fundamental-objective hierar-chy and means-objective network.

Step 4. Extract the attributes for evaluating ERPsystems from the structure of objectives.

Step 5. Filter out unqualied vendors by askingspecic questions, which are formulatedaccording to the system requirements.

Step 6. Evaluate the ERP systems using the AHPmethod.

Step 7. Discuss the results and make the nalstudy, a systematic procedure is proposed toconstruct the objective structure taking intoaccount company strategies and thus extract theassociated attributes for evaluating ERP systems.This study uses the analytical framework of AHPto synthesize decision makers tangible and intan-researchstructurextract the proper criteria for evaluating thent of the companys requirements. Littlehas addressed the issue of objectives for evaluating ERP systems. In thisstructurehow to erelating to the companys strategies and

explainon the elaboration of some commonfor ERP evaluation. However, it did nothow to construct a specic objectiveque andfocusedrk based on the Nominal Group Techni-AHP to select an ERP system. His studyAHP mframewo. Lai et al. (1999) conducted a case studya multimedia authoring system using theethod. Teltumbde (2000) proposed aweightsselectionweights of attributes and applied theseto a goal programming model for ISrelativeC.-C. Wei et al. / Int. J. Proddecision.Fig. 1 shows a owchart for the ERP selectionprocess. The details of each step are presentedbelow.

3.1. Form a project team and collect information on

ERP systems

The rst step is to form a project team thatconsists of decision makers, functional experts andsenior representatives of user departments. Theparticipation and support of top managers sig-nicantly inuences the success of ERP adoption(Ptak, 2000). A wide range of informationconcerning ERP vendors and systems should beobtained from professional magazines, exhibi-tions, yearbooks, the Internet, and other sourcesto ensure that feasible systems are not overlooked.

3.2. Identify the ERP system characteristics

Different companies may adopt an ERP systemfor completely different reasons, including techni-cal and business reasons. The initial rationale foradopting an ERP system inuences problemdenition, methods of achieving goals, and othersubsequent activities. To ensure the processprogresses smoothly from the start, the projectteam has to analyze the ERP selection problem byidentifying decision elements, including the stake-holders, number of alternatives, project objectives,project risks, and other concerns. The project teamcan clarify the complex situation, sketch theproblem and develop an initial plan. The phasealso highlights in advance some difculties innding solutions and supports owing to organiza-tional limitations and resource constraints.

3.3. Construct the structure of objectives

Structuring the objectives involves organizingthem so that the project team can describe in detailwhat a company wants to achieve, and thenincorporate these objectives appropriately intothe decision model. In the rst place, the projectteam should dene the ERP system scope bycompany policy, business attribute, industry en-

Economics 96 (2005) 4762 49vironment, and the goals of the project.

ARTICLE IN PRESS

t team

matio

cteris

obje

ntal- a

uctionConstitute an ERP projec

Collect the relevant infor

Identify the project chara

Discuss and construct the

Distinguish the fundame

C.-C. Wei et al. / Int. J. Prod50Before constructing a detailed objective frame-work, it is necessary to ensure that the strategicobjective scope is appropriate for the projectcharacteristics. The strategic objectives offer asolid basis for decision-making and a stablereference point for ill-structured decision situa-tions. The strategic objectives guide the ultimategoals that the project team should strive toachieve; thus they also serve as the mechanism toharmonize the opinions of different individualswithin the project team.

Establish fundamental-objective hierarchy

Acceptable to senior management?

Extract the attributes for evaluating ERP systems

Form the AHP model

Evaluate the ERP systems by t

Analyze and discuss the result

Make the final decision

Yes

No

Fig. 1. Comprehensive ERP sysn about ERP systems

tics

ctives of project

nd means-objectives

Economics 96 (2005) 4762All objectives derived from the strategic objec-tive scope will be structured systematically. It iscritical to distinguish fundamental-objectives frommeans-objectives in the objective developmentprocess. Fundamental-objectives are those thatare important because they reect what thedecision makers really want to accomplish. Mean-while, means-objectives are those which help thefulllment of other objectives (Clemen, 1996).The fundamental-objectives are organized into a

hierarchy and indicate directions in which the

Establish means-objective network

Develop the attribute details, screening and evaluation guidance

Screen out the unqualified ERP systems

he AHP method

s

Yes

No

Acceptable to senior management?

tem selection framework.

ARTICLE IN PRESS

uctionproject team should strive to perform better. Twomethods can be used to establish the hierarchy ofERP system fundamental-objectives, namely top-down decomposition and bottom-up synthesis. Inthe top-down decomposition procedure, the pro-ject team discusses What do you mean by thatupper-level objective? The answers reveal thelower-level fundamental-objectives, which explainthe meanings of the upper-level objective. Alter-natively, managers can start from a lower-levelobjective by asking, Of what more generalobjective is this aspect? to nd a more generalobjective and move upwards via the bottom-upsynthesis procedure. The upper levels in thehierarchy refer to more general objectives andthe lower levels contain important elaborations ofthe upper objectives. When organizing the hier-archy of fundamental-objectives, the project teammust keep in mind to pay attention to thelimitations of decision elements and the alterna-tion of business environment at any time.Means-objectives are organized into networks

(Clemen, 1996). The project team can create ameans-objective apart from the fundamental-ob-jectives by asking, How could you achieve this?The answers to this question identify the corre-sponding means-objectives and describe the lin-kages among them. Alternatively, the project teamlinks a means-objective toward the correspondingfundamental-objective by asking the questionWhy is that important? Having formulatedthese means-objectives, the project team canensure specic ways of accomplishing the funda-mental-objectives. Additionally, the team cannarrow the set of ERP candidates by examiningthese means-objectives and also develop detailedattribute specications to assess the ERP systems.

3.4. Extract the attributes used for evaluation

After creating the structure of objectives, theproject team can derive the attributes pertinent toevaluating each ERP system. Both quantitativeand qualitative attributes that satisfy the strategiesand goals of the company should be involved.Ideally, the team should develop its own structureof critical objectives and select appropriate mea-

C.-C. Wei et al. / Int. J. Prodsurable attributes to indicate the degree to whichthe corresponding objective is achieved, based onthe business environment and requirements.Therefore, the selected attributes will be consistentwith the objective framework, guided by thecompany strategy.The project team should iteratively examine and

modify the set of selected attributes so that theyare complete, decomposable, nonredundant, mea-surable, and minimal (Keeney and Raiffa, 1993).Then, these attributes will be used as the basis ofthe AHP model.

3.5. Screen the unqualified ERP systems

Numerous alternatives are collected initially,and hence a ltering mechanism is required toshorten the list of ERP candidates. The detailedcharacteristics desired by the company for theERP system are developed over many meetings.Next, suitable characteristics are transferred tospecic requirements to form a questionnaire orchecklist of system specications. Simultaneously,examining the network of means-objectives canhelp to scrutinize system specications and ensurethat these requirements are consistent with corpo-rate objectives. The listed vendors are requested toprovide information in response to the specicquestions included in the questionnaire. Theproject team then assesses this information toeliminate the obviously unqualied vendors.

3.6. Evaluate the ERP systems by using the AHP

method

The AHP is a multi-attribute evaluation methodthat involves three phases: decomposition, com-parative judgments, and synthesis of priorities(Saaty, 1980). In the decomposition phase, theproject team can explicitly develop the AHPhierarchy model from the fundamental-objectivehierarchy as mentioned above. In the secondphase, each decision maker utilizes paired compar-isons for the attributes and alternatives to extractjudgment matrices with a nine-point scale at eachlevel. In the third phase, the paired comparisonprocess is repeated for each attribute in thealternative prioritization problem based on the

Economics 96 (2005) 4762 51largest eigen-value method. Finally, the relative

departments, with at least 5 years experience in

natiregurevi

T

inte

syst

ARTICLE IN PRESS

uctionthe company and expertise in their particularelds, were also chosen to participate in the projectteam. To encourage employee engagement andsupport, the project team held several promotionalworkshops. These meetings produced numerousvaluable recommendations, to which the projectteam responded during the project implementationimportance of attributes and the global priority ofalternatives can be obtained by aggregating theweights over the hierarchy. Hence, AHP canaccelerate the development of a consensusamongst multiple decision makers in ERP systemselection process.

4. Practical example

The proposed framework was applied to ERPsystem selection at an electronics company inHsinchu science park, Taiwan. This companydesigns and manufactures modular microwavecommunication systems and exports them to theUSA, Europe, and Mainland China. On the onehand, the fragmented modules of the existing ERPsystem limited the companys operating efciency,caused much duplication of effort, and confusedthe business processes. On the other hand, thesales cycle of export and the need to maintain goodcustomer service in the highly dynamic businessenvironment put increasingly heavy pressure onthe company. In order to maintain its competi-tiveness, the top management announced thelaunch of a series of projects including theadoption of an ERP system and the re-engineeringof the complex business process to enhance theeffectiveness of its global logistics and providequick responses to customer demands.

4.1. Identify the ERP system characteristics

A steering committee of three senior managerswas formed, including the General Manager andsenior MIS and Purchasing managers, with theresponsibility to formulate the project plan,integrate project resources, and select a suitableERP system. Representatives of different user

C.-C. Wei et al. / Int. J. Prod52to reduce resistance to the project.The process of constructing the objectivestructure of ERP system selection was bothdialectic and analytic, and thus captured themental structure of decision makers. We recom-mended to rst scope out the important strategicobjectives. After discussing with the project team,we gathered the following results:

(1) To satisfy business strategy: to satisfy theindustrial characteristics and the businessgoals and adapt to dynamic business environ-ment.

(2) To enhance business process performance: tointegrate business systems and procedures andenhance information transparency.

(3) To improve operations quality and efciency:to standardize and simplify operations ow,improve quality and reduce lead times.

(4) To shorten turn-around time to the customer:to efciently analyze customer informationfrom various markets and quickly respond tovarious customer demands.

(5) To support globalization development: tosupport business operations worldwide.

Structured interviews and discussions were usedto elicit the structure of objectives from themembers of the project team during several meet-4.2.ingsl legal and tax systems. Initially, 20 ERPems were proposed by various vendors.

Organize the structure of objectivesowilocavered the invitations to various vendors. Toprnational ERP systems were not consideredng to a budget cap and the need to supporttiondelit was preserved.he project team widely collected the informa-on ERP systems and vendors and activelythemenves, and other major concerns at the rstlar project team meeting. This process wasewed iteratively in later meetings to ensure thatexibility to changes in the business environ-tion

P implementation, the project scope, organiza-al strengths and weaknesses, potential alter-To obtain a clear understanding of the crucialelements involved in the decision, we suggestedthat the project team should discuss the goals ofER

Economics 96 (2005) 4762. Fig. 2 displays the fundamental-objective

ARTICLE IN PRESS

inimizst

inimizpleme

uctionMco

Mim

C.-C. Wei et al. / Int. J. Prodhierarchy. The ultimate goal is to select the mostsuitable ERP system. This is divided into twolower-level objective sets, namely choosing themost appropriate ERP system and choosing thebest ERP vendor.We led the team members to discuss What does

choosing the most appropriate ERP system

Selecting the most suitable ERP system

Choosing the best ERP vendor

Choosing the most appropriate ERP system

Having system f

Having interfaceoperatio

Having function

Having reliabili

Having reputatio

Supplyiservice

Providintechnica

Fig. 2. Fundamental-obing total

ing ntation time

Maintenance costs

Price

Consultant expenses

Infrastructure costs

Economics 96 (2005) 4762 53mean? Choosing the most appropriate ERPsystem, meant selecting an ERP system that canminimize total costs and implementation time, hascomplete functionality, user-friendly interface andoperations, and satises system exibility andreliability. On the other hand, the objectives ofselecting the best ERP vendor were those related

Security

Function-fitness

Module completion

excellent lexibility

user-friendly and ns

complete ality

high system ty

good n

ng ongoing

g good l capability

Upgrade ability

Ease of integration

Ease of in-house development

Ease of operation

Ease of learning

Stability

Recovery ability

Financial condition

Scale of vendor

Market share

R&D capability

Warranties

Technical support capability

Implementation ability

Training service

Consultant service

Service speed

jective hierarchy.

ARTICLE IN PRESS

uctionto vendor features, including excellent reputation,technical capability, and ongoing service. Fig. 2shows the two objective sets that constitute thethird level of the fundamental-objective hierarchy.For example, we further asked What doesminimizing total costs mean? to drill down inthe objective of total cost minimization and foundthat the answer lay in minimizing system price,maintenance expenses, consultant expenses, andinfrastructure costs. Similarly, the top-down de-composition method found that the other objec-tives in level 3 must also be decomposed toestablish their practical meanings. Level 4 revealsthe details that can be used to compare theperformance of alternatives.To verify the consistency of the fundamental-

objective hierarchy, we can simultaneously ask theproject team to synthesize the lower-level objec-tives into more general objectives and thus renethe hierarchical structure previously derived fromthe top-down decomposition method. Indeed, thisprocess was iterative and the structure was notunique.The means-objective networks for system and

vendor factors were formulated simultaneously asshown in Figs. 3 and 4, respectively. The projectteam started from the bottom objectives of thefundamental-objective hierarchy (Fig. 2) by askingthe question, How can this objective beachieved? to help the project team to identifythe means-objectives and establish links amongthem. For example, in Fig. 3, the answers to thequestion, How can a lower price for the ERPsystem be achieved? were minimizing the degreeof customization and reducing the price ofvendors quotation. That is, if the team couldprovide clear requirements and specications tothe vendor, then customization could be controlledand the price thus lowered. Furthermore, in orderto have clear requirements, business process re-engineering ought to be done rst. In additional,as illustrated in Fig. 4, the project team concernedthe situation in which an ERP vendor may be toolarge to give sufcient supports to a mid-rangecustomer or an ERP vendor may be too small tosurvive and provide consistent service. Thus theproject team needed to collect and analyze the

C.-C. Wei et al. / Int. J. Prod54ERP market data and the vendor provideddata to ensure that the size of the vendor matcheswell with the size of the company. Following asimilar approach, we systematically elaborated thecauseeffect relationships among all the means-objectives into a complete network. Similarly, theproject team incorporated other means-objectivesinto the network in a logical way.Also, we encouraged the team to repeatedly

examine all means-objective linkages in order toconrm every relationship was reasonable. Theprocess was iterative and diagnostic. For instance,the answer to the question Why are clearrequirements important? was that providing clearrequirements would enable the project team toreduce the degree of customization, produceprecise specications, and conrm all requiredmodules by redening and repositioning thebusiness processes. Furthermore, clear require-ments would also help to identify the systemfunctions and clarify the gap between the ERPsystem and the existing operations procedure. Bythis means, the company would be able to reduceproject cost, assess system suitability, and proposecomplete functional requirements.

4.3. Extract the attributes for evaluating ERP

systems

It may be impractical to make paired compar-isons among the ERP systems with respect to everydetailed dimension in level 4 of the fundamental-objective hierarchy in Fig. 2. The difculty arisesbecause too many attributes lead to numerouspaired comparisons in AHP and cause an inef-cient process. Representatives from different userdepartments in the project team were divided intoresearch groups to gather and evaluate the ERPsystem data based on their specialities and jobresponsibility. For example, the Finance Depart-ment and Purchase Department members joined acost research group to provide nancial data,and the MIS Department members studied thefunctionality, exibility, and reliability of eachERP system as well as the technical ability of eachvendor. The evaluations from the research groupswere discussed in a full assembly of the projectteam. The three major decision makers then

Economics 96 (2005) 4762combined the suggestions of each research group

ARTICLE IN PRESS

Implementation time

Security

Function fitness

Module completion

Flexibility

User friendliness

Functionality

Reliability

Upgrade ability

Ease of integration

Ease of in-house development

Ease of operation

Ease of learning

Stability

Recovery ability

Total costs

Maintenance costs

Price

Consultant expenses

Infrastructure costs

Degree of customization

Price of vendors quotation

Degree of employee cooperation

On schedule

Infrastructure integration

Satisfies requirements

Globalization

Comparison with existing processes

Parameters setting

Permission management

Provision of a guidebook

Step-by-step command

Graphic interface

Few bugs

Ease of integration with other systems

Platform independence

Common programminglanguage

Online help

Online learning

System maturity

Minimum of system break down

Automatic data recovery

Automatic data backup

Payment

Discount

Number of modules

Provision of consultants

Cooperation with the consultant companies

Clear requirements

Hardware support

Other systems support

Multi-language

Multi-currency

Multi-site

System & database protection

Price of consultants

Fig. 3. Means-objective network (system factors).

C.-C. Wei et al. / Int. J. Production Economics 96 (2005) 4762 55

ARTICLE IN PRESS

Reputation

Financial condition

Scale of vendor

Market share

ERP market data

Vendor provided data

Service

Technical capability

R&D capability

Warranties

Technical support capability

Implement- ation ability

Training service

Consultant service

Service speed

R&D investment

Number of R&D engineers

Engineers experience

R&D technology

Diverse product line

Upgrade technology

Warranty terms

Warranty details

Consultant experience

Number of consultants

Sufficient training time

Adequate training lessons

Clear problem-solving program

Online service

Implementation methodology

Domain knowledge

Technology cooperation

Implementation experience

Financial stability

Long-term financial viability

Scale matching

Fig. 4. Means-objective network (vendor factors).

C.-C. Wei et al. / Int. J. Production Economics 96 (2005) 476256

with their subjective opinions to evaluate thealternatives.The fundamental-objective hierarchy was mod-

ied to generate an AHP hierarchy. This AHPhierarchy is composed of four levels, as illustratedin Fig. 5. Level 1 reveals the strategic objective forselecting the most suitable ERP system. Level 2consists of two main objectives, namely choosingthe most appropriate ERP system and selecting thebest vendor. Level 3 contains the associatedattributes that are used to measure various ERPsystems and vendor, respectively. The bottom levelconsists of the alternative ERP systems.Using the means-objective network (Figs. 3 and

4), we assisted the project team to establish theevaluation criteria and specic requirements. Thisprocess can ensure that everyone follows the samecriteria in the evaluation process consistently.Table 1 presents the detailed description of theattributes in the AHP model (Fig. 5) with theassociated means extracted from the means-objective network.

4.4. Screen the ERP alternatives

Unfavorable alternatives were eliminated bythorough examination of system specicationsand requirements derived from the means-objectivenetwork. Table 2 lists some of the questions. Afterpreliminary screening, three systems A, B, and C,were selected and requested to provide detailedproposals for further consideration. Meanwhile,intensive interviews and working meetings werescheduled with each vendor. Further, core businessworkows and special requirements of the com-pany were assessed by running demo scenarios andexamining the capacity of each system to fulllthese key demands. Finally, user representativesconducted unit tests related to system feasibility.

4.5. Evaluate the alternatives using AHP and select

a suitable ERP system

Following the AHP methodology, paired com-parisons of the alternatives on each attribute and

ARTICLE IN PRESS

ImplTime

Func

Tota

User

Flex

Reli

Repu

Techcapa

Serv

C.-C. Wei et al. / Int. J. Production Economics 96 (2005) 4762 57ERPsystemselection

Systemfactors

Vendor factors Fig. 5. AHP hementation

tionality

l costs

friendliness

ibility

ability

tation

nical bility

ice

System A

System B

System C ierarchy.

ARTICLE IN PRESS

s

uctionTable 1

Attribute details

Attributes Evaluation item

C.-C. Wei et al. / Int. J. Prod58the inter-attribute relative importance were madeand converted to a numerical scale of 19. Thesoftware Expert Choice was then used to deter-mine the normalized weights and synthesize the

System software factors Total costs 1. Price

2. Maintenance cost

3. Consultant expen

4. Infrastructure cos

Implementation time

Functionality 1. Module completio

2. Function tness

3. Security

User friendliness 1. Ease of operation

2. Ease of learning

Flexibility 1. Upgrade ability

2. Ease of integratio

3. Ease of in-house d

Reliability 1. Stability

2. Recovery ability

Vendor factors Reputation 1. Scale of vendor

2. Financial conditio

3. Market share

Technical capability 1. R&D ability

2. Technical support

3. Implementation a

Service 1. Warranties

2. Consultant service

3. Training service

4. Service speedMeans

Economics 96 (2005) 4762results. Table 3 lists the inter-attribute pairedcomparison matrices of decision maker 1 forsystem software and vendor attributes. Therelative weights of each attribute for all decision

1. Limited project budget

s 2. Limited annual maintenance budget

ses 3. Limited infrastructure budget

ts

1. 69 months

2. Project management ability

n 1. Availability of necessary modules

2. Parameter setting

3. High function-tness

4. Multi-currency, multi-language, and multi-site

5. Permission management

6. Database protection

1. Graphic interface

2. Step-by-step command

3. Provision of a guidebook

4. Online learning

5. Online help

1. Common programming language

n 2. Platform independence

evelopment 3. Ease of integration with other IS

1. Automatic data recovery

2. Automatic data backup

1. Scale matching

n 2. Financial stability

3. Long-term nancial viability

4. Provision of reference sites

1. Good upgrade service

capability 2. Diverse product line

bility 3. Good implementation experience

4. Ease of implementation

5. Adequate number of engineers

6. Cooperation with other partners

7. Domain knowledge

1. Warranty details

2. Adequate number of experienced consultants

3. Complete training lessons

4. Good problem-solving program

5. Online service

ARTICLE IN PRESS

the p

een c

plex,

or is

uctionTable 2

Examples of screening questions

Items Questions

Cost vs. budget 1. What are the total costs of

2. If a discrepancy exists betw

Complexity 1. Is the ERP system too com

2. Does it t our requirements

C.-C. Wei et al. / Int. J. Prodmakers are listed in Table 4. Table 5 presents theevaluation results of all decision makers, and thelast column of this table indicates the overallpriority of the three ERP systems using thegeometric mean method. As shown in Table 5,system A was the best choice for the company.Throughout the evaluation process, the consis-

tency index (CI) and consistency ratio (CR) ofeach decision makers paired comparison matrixshould be less than the threshold value 0.1 (Saaty,

Requirement coverage 1. Do the system and modules mee

Flexibility 1. Is the technology exible and lon

Fundamentals 1. What database and hardware ca

Information technology 1. Does the vendor provide other in

Manufacturing Execution System (

Management (CRM), and Electron

2. Is the system integrated with tho

Vendor size 1. Does the size of the vendor matc

2. Is the vendor too big to pay atte

consistent service?

Domain knowledge 1. What are the target domain and

2. Do they correspond with the com

Implementation methodology 1. What is the implementation met

2. Is it feasible and simple?

Maintain service 1. Who supports upgrades and mai

2. Does the vendor have any local

Consultant service 1. Does the vendor provide consult

2. Does the vendor co-operate with

Financial considerations 1. What was the nancial performa

2. What are its current nancial fo

3. Does the vendor have any ventur

For example, is this vendor deep in

income substantially dropped in a sroject?

ost and budget then is it acceptable?

too simple or just right?

it overqualied?

Economics 96 (2005) 4762 591980) to ensure that the decision maker wasconsistent in assigning paired comparisons. Other-wise the decision maker may need to reconsider hisevaluation.According to Table 4, the decision makers were

fairly consistent in ranking the attributes. Thefunctionality of the ERP system was ranked rstamong system factors, followed by system ex-ibility, implementation time, total cost, user-friendliness, and system reliability. As for vendor

t all our requirements?

g lasting?

n support the system?

formation systems, such as, Supply Chain Management (SCM),

MES), Data Warehousing (DW), Customer Relationship

ic Commerce (EC)?

se of other partners?

h that of our company?

ntion to us? Or is the vendor too small to survive and provide

market of the provider?

panys needs of the purchasing company?

hodology?

ntenance? The software supplier or the reseller?

service point or branch company?

ing services?

other consultant companies?

nce of the vendor in the past 2 years?

recasts?

e capital investors or show any signs of potential nancial crisis?

debt? Is this vendor a candidate for takeover by another? Has its

hort time?

ARTICLE IN PRESS

F

1/

1/

1

1/

1/

uctionTable 3

Paired comparison judgment matrices (decision maker 1)

Total costs Implementation time

Total costs 1 1/3

Implementation time 3 1

Functionality 3 4

User-friendliness 1 1/3

Flexibility 5 3

C.-C. Wei et al. / Int. J. Prod60selection, all decision makers agreed that technicalability of the vendor was the most importantfactor. Vendor service ability and reputationfollowed as the second and third most importantconsideration.As shown in Table 5, decision makers 1 and 3

preferred System A while decision maker 2preferred System B. However, the difference inthe preference of decision maker 2 for Systems A

Reliability 1/4 1/6 1/

Reputation

Reputation 1

Technical capability 7 1 3

Service 5 1/3 1

Table 4

Relative weights of attributes

Attributes

System factors Total costs

Implementation time

Functionality

User-friendliness

Flexibility

Reliability

Vendor factors Reputation

Technical capability

Service

Table 5

Results of AHP analysis

Alternatives DM 1 DM 2 DM 3 Geometric

mean

System A 0.429 (1) 0.378 (2) 0.449 (1) 0.410 (1)

System B 0.349 (2) 0.381 (1) 0.360 (2) 0.356 (2)

System C 0.224 (3) 0.242 (3) 0.191 (3) 0.210 (3)unctionality User-friendliness Flexibility Reliability

3 1 1/5 4

4 3 1/3 6

7 3 9

7 1 1/5 3

3 5 1 7

9 1/3 1/7 1

Technical ability Service

1/7 1/5

Economics 96 (2005) 4762and B was minimal. The project team thusachieved sufcient agreement to choose System Afor implementation.

5. Discussion

To avoid being constrained by the existingstructure of objectives, the project team shouldask the questions and discuss the answers in anyorder during constructing the fundamental-objec-tive hierarchy and means-objective network(Clemen, 1996). It is important to rene thestructure of objectives iteratively to ensure thatthe decision situation has been examined fully anddifferent perspectives have been considered. Asshown in Fig. 1, during each round of discussion,the team was given an opportunity to review and

DM 1 DM 2 DM 3

0.063 (4) 0.170 (3) 0.119 (4)

0.138 (3) 0.170 (3) 0.128 (3)

0.457 (1) 0.351 (1) 0.420 (1)

0.058 (5) 0.057 (5) 0.048 (5)

0.257 (2) 0.208 (2) 0.247 (2)

0.028 (6) 0.042 (6) 0.039 (6)

0.072 (3) 0.094 (3) 0.072 (3)

0.649 (1) 0.627 (1) 0.649 (1)

0.279 (2) 0.280 (2) 0.279 (2)

strengths, and limitations of the AHP method.

specications. These objectives also indicatehow outcomes should be measured and what

(3)

(4)

Ref

Badr

g

J

Chie

n

T

Clem

t

ARTICLE IN PRESS

uctionDuring the development process, consistencychecks were conducted and the decision makersin some cases were asked to provide reasons anddetailed explications to justify and rene theirassessments.In practice, it seems to be time consuming

to establish an objective framework in the rstplace. Very often, people rush into the stage ofcomparing available alternatives and conductingthe system demonstrations and tests beforethey identify the right problem through needsassessment and objective construction. However,people involved in the project team agreed thatthe entire process enabled detailed review ofthis important project and facilitated theclarication of problems to reach consensusin group decision process. Therefore, thedecision quality was increased while actuallysaving a lot of time in possibly revising thedecisions and projects. In this case, it took about2 weeks in total to construct and rene theobjective structure.

6. Conclusion

This study presents a comprehensive frameworkfor selecting a suitable ERP system based on anAHP-based decision analysis process. The pro-posed procedure allows a company to identify theelements of ERP system selection and formulatethe fundamental-objective hierarchy and means-objective network. The pertinent attributes forevaluating a variety of ERP systems and vendorscan be derived according to the structure ofrevise the hierarchy of fundamental-objectivesand the network of means-objectives to reach aconsensus on ERP system selection.The precision with which decision makers could

provide a paired comparison was limited by theirknowledge, experience, and even cognitive biases,as well as by the complexity of the ERP systemselection problem (Chien et al., 2002). Thus, toavoid inconsistency among semantic descriptionsand score assignments to the attributes, we trainedthe decision makers to understand the details,

C.-C. Wei et al. / Int. J. Prodobjectives.control software. IIE Solutions 27, 1216.

Hong, K.K., Kim, Y.G., 2002. The critical success factors for

ERP implementation: An organizational t perspective.Hick

reIusiness Review 61 (1), 118125.

n, C.F., Chen, S.L., Lin, Y.S., 2002. Using Bayesian

etwork for fault location on distribution feeder. IEEE

ransactions on Power Delivery 17, 785793.

en, R.T., 1996. Making Hard Decisions: An Introduction

o Decision Analysis. Duxbury Press, Pacic Grove.

s, D.A., Stecke, K.E., 1995. The ERP maze: Enterprise

source planning and other production and inventoryBuss

Bphase, but also mitigate the resistance andinvisible costs in the implementation stage.

erences

i, M.A., Davis, D., Davis, D., 2001. A comprehensive 01

oal programming model for project selection. International

ournal of Project Management 19, 243252.

, M.D.J., 1983. How to rank computer projects. Harvardamong multiple decision makers.The approach systematically assesses corpo-rate attributes and guidance based on thecompany goals and strategic development. Itcan not only reduce costs during the selectionkey points should be considered in the decisionprocess.The approach is exible enough to incorporateextra attributes or decision makers in theevaluation. Notably, the proposed frameworkcan accelerate the reaching of consensusThe proposed comprehensive ERP system selec-tion framework has the following advantages:

(1) It ensures that the structure of objectives isconsistent with corporate goals and strategies.The project team can understand the relation-ships among different objectives and assesstheir inuence by modeling them to thehierarchical and network structures.

(2) The project team can decompose the complexERP selection problem into simpler and morelogical judgments of the attributes. Particu-larly, knowledge of structure of objectives canhelp the project team to identify the companyrequirements and develop appropriate system

Economics 96 (2005) 4762 61nformation & Management 40, 2540.

Keeney, R.L., Raiffa, H., 1993. Decisions with Multiple

Objectives: Preferences and Value Tradeoffs. Cambridge

University Press, New York.

Kumar, V., Maheshwari, B., Kumar, U., 2002. Enterprise

resource planning systems adoption process: A survey of

Canadian organizations. International Journal of Produc-

tion Research 40, 509523.

Lai, V.S., Trueblood, R.P., Wong, B.K., 1999. Software

selection: A case study of the application of the analytical

hierarchical process to the selection of a multimedia

authoring system. Information & Management 36, 221232.

Lee, J.W., Kim, S.H., 2000. Using analytic network process and

goal programming for interdependent information system

project selection. Computers & Operations Research 27,

367382.

Lucas, H.C., Moore Jr., J.R., 1976. A multiple-criterion scoring

approach to information system project selection. Infor. 14

(1), 112.

Motwani, J., Mirchandani, D., Madan, M., Gunasekaran, A.,

2002. Successful implementation of ERP projects: Evidence

from two case studies. International Journal of Production

Economics 75, 8396.

Ptak, C.A., 2000. ERP Tools, Techniques, and Applications for

Integrating the Supply Chain. St. Lucie Press, New York.

Saaty, T.L., 1980. The Analytic Hierarchy Process. McGraw-

Hill, New York.

Santhanam, R., Kyparisis, G.J., 1995. A multiple criteria

decision model for information system project selection.

Computers & Operations Research 22 (8), 807818.

Santhanam, R., Kyparisis, G.J., 1996. A decision model for

interdependent information system project selection. Eur-

opean Journal of Operational Research 89, 380399.

Sarkis, J., Sundarraj, R.P., 2000. Factors for strategic evalua-

tion of enterprise information technologies. International

Journal of Physical Distribution & Logistics Management

30 (3/4), 196220.

Schniederjans, M.J., Wilson, R.L., 1991. Using the analytic

hierarchy process and goal programming for information

system project selection. Information & Management 20,

333342.

Teltumbde, A., 2000. A framework of evaluating ERP projects.

International Journal of Production Research 38, 45074520.

Yao, Y., He, H.C., 2000. Data warehousing and the internets

impact on ERP. IT Pro, March/ April, 3741.

Yusuf, Y., Gunasekaran, A., Abthorpe, M.S., 2004. Enterprise

information systems project implementation: A case study

of ERP in Rolls-Royce. Journal of Production Economics

87, 251266.

ARTICLE IN PRESS

C.-C. Wei et al. / Int. J. Production Economics 96 (2005) 476262

An AHP-based approach to ERP system selectionIntroductionSelection method reviewProcedure for selecting a suitable ERP systemForm a project team and collect information on ERP systemsIdentify the ERP system characteristicsConstruct the structure of objectivesExtract the attributes used for evaluationScreen the unqualified ERP systemsEvaluate the ERP systems by using the AHP method

Practical exampleIdentify the ERP system characteristicsOrganize the structure of objectivesExtract the attributes for evaluating ERP systemsScreen the ERP alternativesEvaluate the alternatives using AHP and select a suitable ERP system

DiscussionConclusionReferences