Master Thesis
Identifying the critical success factors for a “vanilla” ERP implementation in SMEs.
Jeroen Wolters
S2594293
H.F. van Riellaan 41B
3571WB Utrecht
J.o.wolters@student.rug.nl
University of Groningen
Faculty of Economics and Business
MSc BA Change Management
28-03-2016
Supervisor: dr. I. Maris-de Bresser
Co-supervisor: dr. U.Y. Eseryel
Co-assessor: dr. C. Reezigt
Word count:13.785
Abstract
So far implementation strategies and in particular a “vanilla” implementation have not been taken into account while identifying Critical Success Factors for ERP implementations in Small to Medium Sized enterprises. This research empirically tests the influence of a “vanilla” implementation on success factors by collecting data from 216 firms using a questionnaire. The empirical evidence shows that co-operation with supplier, detailed scheduling, fast effect, minimal customization and a legacy system are important factors for implementation success. Motivation system, detailed scheduling and investment plan are more important in “vanilla” implementations and in contrast to these results pre-implementation analysis has the opposite effect. In order to increase generalizability, these factors should be tested among other nationalities to take into account cultural factors. Additionally, future research could investigate additional items for success to increase the measure for success. This research increases the understanding of critical success factors and shows the importance of different implementation strategies. Managers can use the identified factors as a guideline during implementations to increase implementation success.
Table of Content
Abstract ... 1
Introduction ... 1
Literature Review and Background ... 3
Implementation Success ... 3
Critical Success Factors... 3
Implementation Strategies ... 9
Methodology ... 16
Factor Analysis and Reliability Testing ... 16
Data Collection and Sample Description ... 16
The Measures ... 17
Analysis and Results ... 18
Descriptive Statistics ... 20
Results ... 21
Discussion and Conclusion ... 29
Theoretical Implications ... 31
Implications for Practice ... 32
Limitations and Future Research Directions ... 32
References ... 33
1
Introduction
Enterprise Resource Planning (ERP) systems are IT innovation that enhances organizational performances through connectivity (Somers and Nelson, 2001). These systems are described as information systems, aimed at processing and facilitating real time organizational transactions (O’Leary, 2000). There is an enormous growth of ERP implementations since 2006 and the reason behind this is that these systems provide integrated business computing solution and improve an organization’s competitiveness (Jacobson, Shepherd, D’Aquila and Carter, 2007). This is done by the integration of data and applications, the replacement of old and outdated systems, cost advantages and adoption of best practices in organizational processes (Brown and Vessey, 1999; Jacobson, Shepherd, D’Aquila, and Carter, 2007). They are designed to solve fragmentation of information within organizations and integrate all the information flowing within a company (Davenport, 1998).
Recent studies showed that the enterprises perceive difficulties with achieving the benefits from the implemented ERP systems. Research of Finney and Corbett (2007) on ERP implementations concluded that ERP implementation could lead to failure or even complete abandonment of the system. Al-Mashari (2000) reported that 70% of the ERP implementations do not achieve the initiated benefits and Rao (2000) made the estimation that 96.4% of ERP implementations fail. Reasons for these failures include ERP systems that are incompatible with strategic partners, spending more money than organizations can afford, a system that conflicts with the management style or overwhelming organizational changes (Luo and Strong, 2004).
Consequently, research started focusing on the implementation process of ERP systems and its critical success factors (CSFs) (Ribbers and Schoo, 2002; Soh, Kien and Tay-Yap, 2000). A large number of researchers focused on identifying the main critical factors for the implementation of ERP systems. Factors such as user involvement, management support, scope management and change readiness were identified (Parr, Shanks and Darke 1999; Esteves and Pastor-Collado, 2001). The identification of the critical factors for the implementation of ERP systems has been well defined in the existing literature (Ahmad and Cuenca, 2013). It appears that much of the literature, however, has focused on success factors with very limited or no regard for differences in implementation strategies. Not all implementations are the same and poses the same characteristics which implies different success factors for different implementation strategies.
2
2013). According to Ahmad and Cuenca (2013), the implementation of ERP systems in SMEs has increased during the last decade as a result of more established technology and decreased prices. A growing number of vendors focusses primarily on SMEs or are expanding their business to the SME sector, which is illustrated by the continuous growth of the ERP market within the SME sector (Pinedo-Cuenca, Shaw, Ahmad and Abbas, 2004). Smaller companies have limited resources and are less likely to overcome a failed implementation than larger organizations (Muscatello, Small and Chen, 2003). This makes it extremely important to study factors influencing successful project outcome within SMEs.Parr and Shank (2000) have highlighted three different implementation strategies in order to maximally understand the process of efficient implementations. These three categories are “comprehensive”, described as the most ambitious implementation approach, “vanilla” as the least risky and “middle-road lies in between. These different strategies provide a way of looking at implementations and pose different characteristics during implementation. The influence of these differences in implementation strategies is the starting point of this research.
The most advised strategy for organizations to minimize the risk of ERP implementations according to the software vendors is to use a “vanilla” implementation, where the organization adopts a software package without modifying it (Bancroft, Seip and Sprengel, 1998; Sumner, 2000; Soh and Sia, 2005) This is preferred because customization is seen as expensive and newer software versions would be harder to install for customized systems (Yusuf, Gunasekaran, and Abthorpe, 2004; Nannery, 1999). Each implementation strategy brings particular characteristics and challenges to the implementation. Both the “vanilla” and “comprehensive” can be used to achieve a fit between the organization and system, but each with its own challenges. “Comprehensive” implementations are featured by more costs, higher levels of risks and complexity (Sia and Soh, 2007). These characteristics differ significantly from those of a “vanilla” implementation such as high speed of implementation and significant changes in structures, roles and processes (Nannery, 1999; Sia and Soh, 2007).
3
questions, this study intends to reduce the risks involved with implementing ERP systems in SMEs. The following research questions are posed:“What are the critical success factors for ERP implementations in SMEs?” “Which of these critical success factors apply for a “vanilla” implementation?”.
Literature Review and Background
Implementation Success
The crucial point of this research is to understand successful implementation of ERP systems. Multiple studies examining CSFs do not define success in their research, or it was defined as satisfying planned time and budget. Success is the subjective assessment of the intention of implementers (Maxie Burns, Turnipseed and Riggs, 1991). Soja (2006) adopts the definition of project success by Brown and Vassey (2003) and defines project success as: “An up and running system with agreed-upon requirements delivered within schedule and budget”(p.412). ERP implementation is aimed at improving workflows, better access to information and improved customer satisfaction and needs to be included in the measurement for implementation success (Hong and Kim, 2002; Lapiedra, Alegre and Chiva, 2011).
The distinction is made between process success, correspondence success and expectation success, which is in line with the study of Hong and Kim (2002). Process success is achieved when an IT project is completed within time and budget. Correspondence success means a match between IT systems and the specific planned performances. Expectation success is when an IT system matches the expected benefits. To measure success in this study, success is defined in line with the research of Hong and Kim (2002). Implementation success is the degree of deviation from project goals in terms of expected costs, time, benefits and system performance. Hence cost overrun, schedule overrun, system performance deficit or the failure to achieve the expected benefits, mean lower levels of implementation success. In contrast, on time, on budget, achieving the expected benefits of the system and achieving expected system performance, stand for higher levels of implementation success. Critical Success Factors
4
The process of identifying critical success factors is based on the work of Rockhart (1979). He developed the concept of identifying critical success factors, to make sure that the necessary factors receive attention. These key-jobs must be done exceedingly good for a company to be successful. “Critical success factors thus a limited number of areas in which results, if they are satisfactory, will ensure successful competitive performance for the organization” (Rockhart, 1979. p.85).CSFs are highly effective in helping executives to define their information needs (Rockhart, 1979). CSFs have been applied to many different aspects of information systems and more recently they started focusing on ERP system implementations (Holland, Light and Gibson, 1999; Kale, 2000; Sumner, 1999). Ultimately, this will increase the likelihood of achieving higher success rates, which will result in bigger cost- and time-savings and a higher efficiency and quality in their system (Finney and Corbett, 2007). More attention to the significant CSFs of “vanilla” implementations, results in a higher chance of a successful implementation.
As mentioned previously, many researchers have focused on identifying CSFs for the implementation of ERP systems. These factors could have an influence on the positive outcome of ERP implementation whereas the absence of these factors might lead to problems. Soja (2004) developed a general model of ERP implementation CSFs based on several literature studies. The study done by Soja (2006) has not been empirically tested nor has it taken different implementation strategies into account, but instead reported the subjective opinions of respondents.
Soja (2006) showed differences in CSF for small and large firms. There are reasons to believe that for SMEs the success rates are even lower than for larger companies (Olsen and Sætre, 2007). For smaller organizations it is more essential that the implementation is successful since they don’t have the financial resources to recover from unsuccessful implementations (Muscatello et al., 2003). “Vanilla” implementations are more used in small organizations (Parr and Shank, 2003). Due to the continuous growth of the ERP market within the SME sector (Pinedo-Cuenca, Shaw, Ahmad and Abbas, 2004), successful implementations for SMEs becomes more relevant. SMEs are defined by Schmiemann (2008) as organizations with 1-249 employee.
5
Table 1. The General Model of ERP Implementation Success Factors
Factor Factor description
Implementation Participation
(A1) Project manager The person from the organization who sacrifices most of his working time to implementation duties. (A2) Team composition The implementation team consists of various people having high qualifications and knowledge about the
enterprise.
(A3) Team involvement The project manager and members of the implementation team are strongly involved in the implementation duties.
(A4) Motivation system There is a motivation system rewarding participation in implementation and on-time task delivery.
(A5) Co-operation with supplier Good co-operation with the system supplier who is competent and offers high levels of services. Top management involvement
(B1) Top management support Top management support for the project and the management members' involvement in implementation duties. (B2) Top management awareness Top management awareness regarding the project goals, complexity, labour required, existing limitations,
required capital investment and project inevitability.
(B3) Top management participation Top management participation in the project schedule and goals definition. Project definition and organization
(C1) Linking with strategy Linking the implementation project with organization strategy (implementation as a method of achieving the enterprise's strategic goals).
(C2) Implementation goals The definition of implementation goals, defined in economic terms at the organization-wide level.
(C3) Detailed schedule The definition of detailed implementation scope, plan and schedule with responsibility allocation.
(C4) Pre-implementation analysis Organization analysis and diagnosis prior to the start of implementation, and the creation of the organization functioning model with the integrated system support.
(C5) Organizational change The change in the organization and its business processes.
(C6) Monitoring and feedback Information exchange between the project team and end-users.
(C7) Implementation promotion Information broadcasting about the project by the implementation team members to other employees. (C8) Fast effects The visible, fast, partial, positive results of the implementation.
6
Project status
(D1) Investment plan Formal introduction of the implementation project in the organization hierarchy.
(D2) Project team empowerment The empowerment of the project team members to make decisions and their high position in the organization hierarchy.
(D3) Financial budget The financial resources assured during the implementation.
(D4) Work time schedule The work time assured for the implementation team members (work time schedule).
(D5) IT infrastructure The appropriate IT infrastructure assured for the implementation project. Information systems
(E1) System reliability The ERP system reliability, its user friendliness and fit to the enterprise's needs. (E2) Minimal customization The use of defined patterns and solutions embedded in the system.
(E3) Legacy systems The legacy systems adaptation for the operation in the ERP integrated system environment.
7
Main hypothesis development. The category of success factors related to implementation participation such as team-involvement, has been reported to relate positively to positive change outcomes, such as change acceptance and support for the change (Armenakis and Bedeian, 1999; Holt, Armenakis, Field and Harris, 2007). Team involvement in this study is described as members of the implementation team who are strongly involved in the implementation duties. Kotter and Schlesinger (1979) stressed that to increase acceptance of the change, participation of the members is needed. Therefore, this research proposes that participation will increase the level of success of the implementation. This leads to the following hypothesis:
H1: “Critical success factors with relation to participation are positively associated with implementation success”. In particular: (a) Project team manager (b) Team composition (c) Team involvement and (d) Cooperation with supplier.
The involvement of top management is seen as an indicator for ERP implementation success in the literature. Top management can create an effective awareness under the workers and contribute to success through management commitment and support (Aladwani, 2001). In a survey done by Zairi and Sinclair (1995), leadership was ranked as the biggest facilitator for transformation efforts. Furthermore, top management provides the resources necessary for the implementation (Parr and Shanks, 2003). The support of top management can play a role in settling disputes in organizations and provide clear directions (Zhang, Lee, Huang, Zhang and Huang, 2005). These arguments build a strong case for the involvement of top management in ERP implementation. This leads to the following hypothesis:
H2: “Critical success factors with relation to top-management involvement are positively associated with implementation success”. In particular: (a) Top-management support, (b) Top-management awareness (c) Top-management participation.
8
success, for example cutting costs for training. Cutting costs for training might lead to wrong data entry, and recovery from this. This builds up to a case where higher levels of project definition and organization, would result in a more successful ERP implementation. This leads to the following hypothesis:H3: “Critical success factors with relation to project definition and organization are positively associated with implementation success”. In particular: (a) Linking with strategy, (b) Implementation goals (c) Detailed schedule (d) Pre-implementation analysis (e) Organisational change (f) Monitor and feedback (g) Implementation promotion (h) Fast effect (i) Appropriate training.
Project status is the collective noun of several factors. Project team empowerment is one of these factors and relates to team members who are empowered to make decisions. Parr and Shank (2003) reveal that in ERP implementation organizations run risks such as lack of adequate control over increased responsibilities. The empowerment of lower level employees must always be done during the implementation (Parr and Shank 2003). Without this empowerment, the organization might have inadequate control, because the members with relevant specific knowledge do not have the right to decide.
IT infrastructure also falls under project status in the categorization by Soja (2006). IT infrastructure consists of re-useable and shareable resources, which provides bases for present and future IT applications (Duncan, 1995). IT systems with standard application architecture provide an infrastructure that supports business flexibility for change such as ERP implementations (Parr and Shank, 2003). So the current IT infrastructure has its influence on ERP implementations and can support this.
A third factor of project status is financial budgeting. One of the major treats to project success is the overrun of budgets (Parr and Shank, 2003). When budgets run short, the users were unable to adopt the most preferable processes and therefore adopted inefficient processes. When the financial resources are assured for the implementation project, these inefficient processes will not be adopted and this will influence the implementation positively. These factors jointly contribute to project status and its positive influence on implementation success. This leads to the following hypothesis:
H4:“Critical success factors with relation to project status are positively associated with implementation success”. In particular: (a) Investment plan (b) Project team empowerment (c) Financial budget (d) Work time schedule (e) IT infrastructure.
9
members can apply experiences gained in the past onto the new situation. Learning by experience was already observed as a valid approach (Levitt and March 1988). Based on past experiences there will be a better recognition of problems, better understanding of future problems and use of past experiences to solve the problems (Parr and Shanks 1999). Furthermore, Amburgey, Kelly and Barnett (1993) showed that organizations learn to change by changing.System reliability is another factor of information systems. The reliability is determined by, among other things, its user-friendliness. Zhang et al., (2005) showed in their research on implementations in China, that system quality was one of the requirements for implementations. This system quality consisted of user-friendliness, but also system reliability. Based on these arguments, a case is built for the positive influence of information systems on implementation success. This leads to the following hypothesis:
H5: “Critical success factors with relation to Information system are positively associated with implementation success”. In particular: (a) System reliability (b) Legacy system (c) Implementation experience.
Implementation Strategies
In the literature multiple theories have been applied to investigate ERP systems and organizational change. Boudreau and Robey (1999) listed several theories to describe the ERP system implementations and found that multiple theoretical lenses were applicable for this phenomenon. Evolutional theory is applicable because it refers to changes in structural forms of population entities through a cycle of variation, selection and retention (Van der Ven and Poole, 1995). Teleological theory have been applied with the assumption that organizations have constructed an envisioned end state and takes actions to reach this while monitoring the progress (Braddon-Mitchell and Jackson, 1997). Paré and Elam (1997) have applied this theory to the implementation of an information system in health care organizations. Theories of organizational learning are specifically appealing because they represent the knowledge base members of the organization must learn in implementations and that these members need to change their task-oriented view (Boudreau and Robey, 1999). This is important because the implementation of an ERP system crosses organizational boundaries. The focus of these theories is rather internal and aim at the changes the organizations go through, instead of changes in its surroundings. This makes them applicable on “vanilla” implementations.
10
reflection of rules, norms and procedures). Any ERP implementation requires a fit between the system that is implemented and the processes in the organization that the system supports (Robey, Ross and Boudeau, 2002).ERP implementation is viewed as a generic concept in most research on successful ERP implementation. However, Parr and Shanks (2000) argue that the concept of ERP implementation is not a generic concept, but rather a taxonomy of implementing categories. ERP implementations fall into three archetypal categories; “comprehensive”, “middle-road” and “vanilla”. The main difference between these three categories is based on how each copes with misalignments between system and organization. Where “vanilla” or process customization achieved a fit by changing the processes in the organization, “comprehensive” or technical customization means altering the system to achieve a fit. Middle road lies between these two extremes and a continuum arises with these three categories, see Figure 1.
Figure 1. Conceptual Overview ERP Implementation Strategies
"Vanilla" “Comprehensive”
0% technical customization 100% technical customization
100% process customization 0% process customization
The first category “Comprehensive” is characterised as the most ambitious implementation approach (100% technical customization, 0% process customization). This method is favoured by multinationals and involves the effort to implement all the modules of the ERP package, without business process reengineering (Lee, Siau and Hong, 2003). Companies have three possible options to alter the system and reduce misalignment, namely module selection, table configuration and code modification (Davenport, 1998). It’s beyond the scope of this research to describe these options in detail.
11
The “Middle-road” category lies between the other two categories (50% technical customization, 50% process customization). By modifying both, a fit can be achieved. Similar to “comprehensive” this also entails some risk of failure because it requires process and system changes.The third category of ERP implementations is the “vanilla” implementation approach (0% technical customization, 100% process customization) and is the opposite of “comprehensive” implementations. Organizations adopt the package as imposed by the developer without modifying it. It requires organizational adaptation instead of ERP modification (Soh and Sia, 2005) and a fit is achieved by aligning the business processes to the ERP system. Typically, such a system is used by a small number of users, so less than 100 (Parr and Shanks, 2000). “Vanilla” implementation is characterised by high speed of implementation which reduces both risks and implementation time (Nannery, 1999). Furthermore, future upgrades will be better supported and this form of implementation is less costly than the other options (Sia and Soh, 2007).
Grover and Kettinger (1995) define business process as a set of logically related tasks that uses organizational resources to achieve a defined business outcome. This definition points out that business processes consist of several elements, namely the outcome, tasks, relationships between tasks, resources and relationships among the tasks and resources. When the organization chooses to alter its business processes, these elements and its relationships are altered and three process customization categories arise (Grover and Kettinger, 1995).
The first category is no change, were only tasks and resources are changed but not the relationship among these elements. For example, the substitution of manual labour for computer systems. The relationship within the business processes stays the same, but resources are changed. The second category is incremental change, describing changes in resources and tasks, but also in the relationships between these elements (Luo and Strong, 2004). This does not imply changes in the nature of the outcome. The third category is radical change, meaning that all the elements of the business process and its measure for performance are radically redesigned. The category of process customization, depends on the misfit between the organization and the ERP system.
Based on the previous described characteristics, it is clear that the different implementing strategies have different consequences. Due to the fundamental differences between these classifications, it is too bluntly to say that these categories have the same critical success factors as previously assumed in the literature. Complexity, speed of implementation, costs, process and role changes all differ between these implementation strategies.
12
the negotiation and interaction between members, managers and users. Hence, “vanilla” implementations require individual and group adaptation, enhances collaboration between the recipients and increases organizational members and group adapting (Soh and Sia, 2005). The ingrained characteristics of “vanilla” implementation members interaction, collaboration and negotiation, will have a positive influence on the relationship between participation and the level of implementation success. With higher levels of “vanilla” implementation these different groups and individuals have to adapt to the new organization. Haines’s (2009) research on system customization showed that user involvement can reduce the need for customization because involvement leads to less resistance and increased user buy-in. This leads to the following hypothesis:H6: “Higher levels of “vanilla” implementations will increase the effect of implementation participants on implementation success”.
The implementation of a “vanilla” ERP system requires the organization to change and adapt by, for example, conscious redesign (Soh and Sia, 2006). Here a strong top management is important in formulating and implementing strategic change (Soh and Sia, 2006; Mintzberg, 1979). The top management of an organization is the “dominant coalition” of individuals and interprets relevant information, identifies opportunities and formulates and implements strategic change (Mintzberg, 1979). This description stresses the task of top management to implement strategic change which is needed to adapt the organization to the new ERP system.
The manageable system-organization misalignments can be reduced but the organization needs to plan adequate funding and time to modify and test the new situation (Soh and Sia, 2005). These actions need to be accomplished by top management, which is in line with the factors underlying the category “top management involvement” such as top management participation in scheduling and goal definition (Soja, 2006). Haines (2009) states that organizational behaviour, structure and processes are determined by the choices of top management, and thus not its members. Members possess the in-depth knowledge of the embedded package structures and program (Soh and Sia, 2005). Yet, because the organization is changing and not the ERP system, the knowledge of its members is less important and it come down to the in-depth knowledge of top management.
Therefore, with higher levels of “vanilla” implementation, the role of top management is more important for implementation success then in a situation where the software is customized. This leads to the following hypothesis:
H7: “Higher levels of “vanilla” implementations will increase the effect of top management involvement on implementation success.”
13
developed system is not custom build for particular organizations (Soh and Sia, 2005). When the ERP system is implemented as a method of achieving organizations goals, and they have formulated implementation goals, this has its effect on implementation success. With higher levels of “vanilla” implementation these goals and the link with its strategy will not be achieved completely as it is with customized ERP systems.Swan, Newell and Robertson (1999) found that these systems, not developed for a particular organization, do not meet all the specific needs of the customer. The disappointing configurations reduce the effect of the link with its strategy and implantation goals on implementation success by not completely meeting the needs of the organization. Furthermore, Soh and Sia (2005) showed that strong justification for package modification was found for systems that needed to be adapted because of the organizations strategy. The more “vanilla” the implementation, the less this will be done and this will influence the relation negatively because not all the mismatches between organization and system will be resolved.
These disappointing configurations due to the “vanilla” implementation will also have its negative influence on the relation between detailed planning and implementation success. These plans will not be accomplished and the scope of the change can also change as a result of not meeting the specific needs of the organization. Tian and Xin (2015), for example, showed that an increase in system scope enables organizations to effectively change their operating environment. So the effect of this factor is also negatively influenced by a “vanilla” implementation.
The next factor in the category project definition and organization is monitor and feedback. This is defined as information exchange between end-users and team members (Soja, 2006). But the users who provide this feedback on the system, need to know that this feedback will be received and acted upon (Fui-Hoon Nah, Lee-Shang Lau and Kuang, 2001). In “vanilla” implementations, this will not happen with regard to the system because it will not be customized although this might be needed. These arguments build a case for the negative influence of higher levels of “vanilla” implementation on the effect of project definition and organization on implementation success. This leads to the following hypothesis:
H8: “Higher levels of “vanilla” implementations will decrease the effect of project definition and organization on implementation success.”
14
for future changes and increase the capabilities of an organization including quick and economic implementation of new applications (Parr and Shank, 1999). Hence, with appropriate IT infrastructure, the misfits with the “vanilla” system will not be present or minimized due to the fact that this appropriate IT infrastructure supports flexibility for changes and supports these changes. This effect is increased by higher levels of “vanilla” implementation, which is characterized with high speed of implementation (Nannery, 1999). “Vanilla” implemented systems also support system updates better then customized systems (Nannery, 1999). So the positive influence of appropriate IT structure on successful implementation, will be increased by higher levels of “vanilla” implementation because of its speed of implementation and its support for future changes.Two other factors in this category are financial budget and time schedule. Financial budget means that if the financial resources are assured during implementation, this would lead to more implementation success. Work time schedule means the time assured for the implementation team members. The “vanilla” implementation is less risky and costly then the implementation of a customized system (Parr and Shanks, 2000). Customization is characterized by high levels of complexity and this results in more difficulties in the planning and estimation of time, resources and costs (Parr and Shanks, 2000). With high levels of “vanilla” this leads to an increase in the relationship between financial budgeting and time schedule on implementation success. With low levels of “vanilla” implementation, complexity and difficulties with planning increases, work is less predictable and this reduces the influence of financial budgeting and work time schedule on implementation success.
Based on these arguments one would expect a positive effect of higher levels of “vanilla” on the relation between project status and implementation success. This leads to the following hypothesis: H9: “Higher levels of “vanilla” implementations will increase the effect of project status on implementation success.”
The last category of factors is the information system with implementation experiences as one of its factors. This describes the experiences organizational members have gained during former implementations (Soja, 2005). Members of the organization push for package modification because they want to minimize the amount of change they have to make (Soh and Sia, 2005). Members have a more negative idea about how the change is implemented (e.g. “vanilla” instead of customization). This is part of the change readiness and this reflects the extent to which individuals are inclined to accept, embrace and adopt a particular plan (Holt et al., 2007). The influence of previous change experience will be reduced for organizations with higher levels of “vanilla” because of this lower acceptance how the change is implemented.
15
fundamental misalignments and these are discovered late in the implementation (Soh and Sia, 2005). With customized systems this misalignment does not occur. Therefore, with higher levels of “vanilla”, the chances of undetected misalignment will be increased. This leads to a negative influence on the effect of system reliability on implementation success. These arguments together build a case for the negative influence of higher levels of “vanilla” implementation on the relation between system reliability on implementation success. This leads to the following hypothesis:H10: “Higher levels of “vanilla” implementations will decrease the effect of information systems on implementation success”.
A conceptual overview of this research is presented in Figure 2. Figure 2. Conceptual Model
H3 + Implementation participation
(A1) Project manager (A2) Team composition (A3) Team involvement (A4) Motivation system (A5) Co-operation with supplier
Top management involvement (B1) Top management support (B2) Top management awareness (B3) Top management participation
Top management involvement (B1) Top management support (B2) Top management awareness (B3) Top management participation
Project definition and organisation (C1) Linking with strategy (C2) Top management awareness (C3) Detailed schedule (C4) Pre-implementation analysis (C5) Organizational change (C6) Monitoring and feedback (C7) Implementation promotion (C8) Fast effects
(C9) Appropriate training
Project definition and organisation (C1) Linking with strategy (C2) Top management awareness (C3) Detailed schedule (C4) Pre-implementation analysis (C5) Organizational change (C6) Monitoring and feedback (C7) Implementation promotion (C8) Fast effects
(C9) Appropriate training
Project status (D1) Investment plan (D2) Project team empowerment (D3) Financial budget (D4) Work time schedule (D5) IT infrastructure
Project status (D1) Investment plan (D2) Project team empowerment (D3) Financial budget (D4) Work time schedule (D5) IT infrastructure
Information system (E1) System reliability (E2) Minimal customization (E3) Legacy system
(E4) Implementation experience
Information system (E1) System reliability (E2) Minimal customization (E3) Legacy system
(E4) Implementation experience
Implementation success Implementation success H1 + H2 + H4 + H5 + Level of “vanilla” implementation Level of “vanilla” implementation H7 + H8 - H9 + H10 -H6 +
16
Methodology
CSFs can be researched using both quantitative and qualitative approaches. Although a qualitative approach provide rich information, researchers have used quantitative approaches to research CSFs (Hwang and Lim, 2013) and this better suits this research purpose. A quantitative approach is chosen when there is a mature literature field, while not all relations among the concepts have been ultimately identified (van Aken, Berends and Van der Bij, 2012). This method fits this research since it converts phenomena into numeric values so that an statistical analysis can be conducted which enables the exploration of causal relations among variables (Gelo, Braakmann and Benetka, 2008). By studying a representative research sample, this research tries to identify relations and provide generalizable statements about CSF in SMEs (Gable, 1994).
Factor Analysis and Reliability Testing
Before the data could be tested for possible relations, a factor analysis was conducted. As commonly agreed, the purpose of a factor analysis is to gain information about the interdependencies between variables and their items (Field, 2013). Items were only included if they fulfilled the following criteria: The measure must have a loading of higher than 0.5 (Song, Bij and Song, 2011). After the factor analysis, reliability analysis have been done for the multi-item constructs and are presented by means of their Cronbach alpha.
Data Collection and Sample Description
The English version of the questionnaire was translated to Dutch since this research solely focusses on Dutch SMEs. A pilot survey was presented in person to one professor and two students to review the translation and survey questions. The pilot study resulted in some revisions and modifications of the questionnaire.
The revised version of questionnaires was distributed to those who fall in a category of two criteria. Firstly, the SME had to have a ERP implementation within the last 5 years. Secondly, the person participating in this research needed to be involved or leading the implementation because the questions of this research focus on factors present during the implementation. So it was critical that the respondent were involved in the implementation. Huber and Powere (1985) state that if one respondent per organization is questioned, the person who is knowledgeable about the issue of interest needs to be identified, in this case knowledgeable about ERP implementation.
17
information of those involved in the implementation. Surprisingly, these involved persons were seldom persons from the management team. Furthermore, the researcher contacted over 54 ERP vendors if they wanted to assist in distributing the questionnaire. This resulted in 326 contact persons involved in implementing ERP systems in SMEs. These persons were contacted by phone to explain the research and encouraged to fill out the questionnaire. The phone call was followed up by an e-mail with additional information and a link to the anonymous survey obtainable via Qualtrics. The data was gathered for a period of four weeks. Per organization only one person filled in the questionnaire and this was checked by analysing IP-addresses.This led to 219 valid respondents (67%) and 194 completed surveys (89%) which is a substantial sample size. The average organization size was 60 employees. The range of organizations, runs from 2 -249 employees with a modus of 15 organization with 20 employees. Organizations in agriculture, machine industry, metal industry, marketing, construction, maritime electro and others are included. The Measures
This study aims at researching three concepts and their relations. All study measures are adopted from well validated measures (See appendix I for measuring items and the sources). One control variable is added based on the literature, namely size.
Dependent variable. The first concept is success of the implementation and is the dependent variable in this research. In order to assess the overall implementation outcome, Hong and Kim (2002) developed a measurement to measure implementation success. As described in the literature review, this concept consists of four partial measures. In this study, ERP implementation success will be measured in terms of deviation from expected project goals such as cost overrun, schedule overrun, system performance deficit, and failure to achieve expected benefits which is in line with the study of Hong and Kim (2002). This research adopted a well validated seven-item Likert-type scale running from (1) extremely disagree to (7) extremely agree, to measure the extent to which respondents disagree with statements about the four items of success. Example item for this concept is: “The ERP project took significantly longer than expected”.
18
Independent variable. The third concepts and the independent variable in this research are the Critical Success Factors, presented in Table 1. The respondents were asked to indicate if these factors were present during the implementation. For each of these CSFs, respondents expressed their answers using a five-item Linkert scale ranging from (1) I extremely disagree to (5) I extremely agree. These are the same scales used in the study of Soja (2006). The questions were altered in comparison to the study of Soja (2006). In their research, experts were asked if they thought particular factors were important during the implementation which is a subjective manner of researching (Soja, 2006). In this research, respondents were asked if these factors were present during the implementation, which is a more objective method. An example item is: “There was good co-operation with the system supplier who is competent and offers high level of services”.
Control variable. Firm size is measured by natural logarithm of the number of employees in the organization and is the control variable in this research (Chandy and Tellis, 2000). Only SMEs ( number of employees < 249) are included in this research. This control variable is added to control for the possible effect of other variables on the dependent variable, which is in line with the research of Premkumar and Roberts (1999) who found that organizational size has its influence on the adoption of IT.
Analysis and Results
Hierarchical linear regression analysis was used to test the hypothesis in this research. The least squares regression method was appropriate for this studies data, its simple conceptual model and gives accurate estimations of the correlation (Crawford, 2006; Natrella, 2010). The purpose of a regression analysis is to determine whether there is a causal relationship between the dependent variable and the independent factors (Field, 2013). Hierarchy testing was done to see if the success factors and interactions helped explaining implementation success more then only the control variable, which is ideally for theory-based hypotheses (Petrocelli, 2003). Furthermore, moderation analysis were conducted. By applying a moderation analysis, the strength or direction of a causal relation is modified. This analysis will show “when” or “for whom” an independent variable more strongly (or weakly) causes a dependent variable (Wu and Zumbo, 2007). This way of analysing provides information , which factors are more important with higher levels of “vanilla” implementations.
19
“vanilla” implementation and implementation success. Multicollinearity diagnostic test were done to test how much the independent variables are inter correlated (Grewal, Cote and Baumgartner, 2004).Critical success factors. Since the presence of the factors was measured on a single item scale, no additional analysis needed to be done to prepare the data for hypothesis testing. To test these factors with the interaction terms, these variables are mean centred to minimize the risk for multicollinearity as recommended by Aiken, West and Reno (1991).
Implementation success. Implementation success is measured by means of four items. To test validity, a principle factor analysis was conducted (Angst and Agarwal, 2009). The principal factor analysis for this construct showed values ranging from .65 to .76 and therefore the construct satisfies the criteria. Consequentially, all items referring to implementation success were used in the analysis. Cronbach alpha values, representing the internal consistency and so reliability, were acceptable for the measures (Field, 2013). The respondents ratings for these four relevant items are added up and divided by the number of items to obtain the composite scale for this variable due to a Cronbach alpha of .68.
Level of customization. The level of customization was measured by five items ranging from strongly disagree to strongly agree (Brooke, 1996). The factor analysis satisfied the criteria and for these five items a Cronbach’s alpha of .68 was measured meaning acceptable reliability. Based on the reliability analysis, a sum variable was computed. Interaction terms were computed to see the effect of higher levels of “vanilla” implementations on the direct relation as recommended by Wu and Zumbo (2008).The final set of scale items and their factor loadings are provided in Table 2. The sum variable was mean centred to test interaction effects and minimize the risk for multicollinearity as recommended by Aiken, West and Reno (1991).
Control variable. The results were controlled for organization size. This was measured on a single-item scale and thus do not require additional analysis.
Table 2
Items and Factor Loadings for Implementation SuccessConstruct Items Means S.D. Loading N
Implementation success Imp_S_(1) 3.78 1.70 0.76 193
Imp_S_(2) 3.27 1.78 0.75 193
Imp_S_(3) 4.89 1.53 0.70 193
Imp_S_(4) 5.11 1.44 0.65 193
Level of Customization Cus_(2) 5.15 1.73 .77 192
Cus_(3) 5.49 1.28 .73 192
Cus_(4) 4.83 1.59 .62 192
Cus_(5) 4.51 1.95 .62 192
Cus_(6) 4.56 1.81 .55 192
20
Descriptive Statistics
Table 3 presents descriptive statistics. The correlation matrix of these factors is presented in appendix II.
Table 3. Descriptive Statistics
Factor Mean S.D N Implementation success 4.26 1.16 193 Level of customization 4.91 1.11 192 Project manager 3.51 1.15 190 Team composition 3.88 .80 188 Team involvement 3.50 .84 189 Motivation system 2.83 .98 189
Co-operation with supplier 3.43 1.07 189
Top management support 1.66 .76 190
Top management awareness 3.62 .89 190 Top management participation 3.67 .89 189
Linking with strategy 3.66 .86 190
Implementation goals 3.51 .92 190
Detailed schedule 3.55 .83 190
Pre-implementation analysis 3,18 .92 190
Organizational change 3.57 .78 190
Monitoring and feedback 3.15 .93 189
Implementation promotion 3.37 .81 189
Fast effects 3.12 .90 190
Appropriate training 3.24 .95 190
Investment plan 4.02 .72 190
Project team empowerment 3.94 .71 190
Financial budget 3.32 .96 190
Work time schedule 3.98 .63 190
IT infrastructure 3.73 .81 189
System reliability 3.52 .92 190
Minimal customization 2.71 1.01 190
Legacy systems 2.98 1.11 190
Implementation experience 3.76 .74 190 S.D. indicates Standard Deviation.
21
Investment plan is the factor most present during ERP implementations in SMEs, followed by work time schedule, project team empowerment and team composition. Top management support is the factor that is the least present during ERP implementations in SMEs. Motivation systems and minimal customization are not much observed. Additional analysis shows that higher levels of “vanilla” lead to higher levels of implementation success (β = .15, p < .05) in SMEs.Results
For all factors the correlation coefficient between the level of factor occurrence and implementation success was calculated. Via this method, the impact of each factor on the successfulness of the project is calculated. These correlation coefficients show the CSFs for the ERP implementation. The level of “vanilla” implementation is added to see if there is a significant (p < .05) influence of the implementation strategy on CSFs used by the different organizations. Therefore the regression coefficient is used to examine the hypothesis (Song et al, 2011). Multicollinearity diagnostic test indicate that there is no multicollinearity problem in the regression models with all Variance Inflation Factors < 10 (Belsley, Kuh and Welsch, 1980). Table 4 reports the results of the hierarchical regressions for implementation success.
22
Table 4. Results from Hierarchical Regression Analyses
Model 1 Model 2 Model 3
Coefficient Estimate Standard Error Coefficient Estimate Standard Error Coefficient Estimate Standard Error
β β β Implementation success 4.25 .133 -1.25 .85 .53 6.03 "Vanilla" .15 ** .08 Project manager -.7 .07 -.08 .9 Team composition .05 .08 .09 .9 Team involvement .04 .11 .05 .12 Motivation system -.02 .10 -.06 .11
Co-operation with supplier .16 ** .09 .15 * .09
Top management support 0.09 .12 -.22 ** .13
Top management awareness .08 .12 .15 .15
Top management participation -.05 .07 -.07 .07
Linking with strategy -.10 .10 -.15 * .11
Implementation goals -.09 .07 -.07 .07
Detailed schedule .18 ** .09 .21 *** .11
Pre-implementation analysis .09 .11 .11 .13
Organizational change -.07 .07 .06 .08
Monitoring and feedback -.02 .10 .02 .11
Implementation promotion -.03 .09 -.05 .10
Fast effects .32 *** .12 .28 ** .13
Appropriate training -.03 .09 -.04 .11
Investment plan .02 .09 .00 .09
Project team empowerment .10 .11 .12 .13
Financial budget .02 .12 .00 .14
Work time schedule .04 .08 .03 .09
23
System reliability .57 *** .10 .57 *** .11
Minimal customization .23 *** .08 .19 *** .09
Legacy systems -.01 .07 .00 .08
Implementation experience .07 .06 .07 .07
Project manager x "Vanilla" -.05 .07
Team composition x "Vanilla" -.04 .10
Team involvement x "Vanilla" -.03 .12
Motivation system x "Vanilla" .16 * .10
Co-operation with supplier x "Vanilla" -.08 .09
Top management support x "Vanilla" -.19 .14
Top management awareness x "Vanilla" .07 .13
Top management participation x "Vanilla" -.06 .08
Linking with strategy x "Vanilla" -.03 .14
Implementation goals x "Vanilla" -.04 .08
Detailed schedule x "Vanilla" .20 *** .09
Pre-implementation analysis x "Vanilla" -.18 * .13
Organizational change x "Vanilla" .02 .09
Monitoring and feedback x "Vanilla" -.07 .12
Implementation promotion x "Vanilla" -.05 .10
Fast effects x "Vanilla" -.02 .12
Appropriate training x "Vanilla" -.02 .11
Investment plan x "Vanilla" .13 * .09
Project team empowerment x "Vanilla" .14 .16
Financial budget x "Vanilla" -.13 .15
Work time schedule x "Vanilla" .-8 .09
IT infrastructure x "Vanilla" .15 .17
System reliability x "Vanilla" .02 .12
Legacy systems x "Vanilla" .01 .08
Implementation experience x "Vanilla" -.04 .07
24
Firm size 0.00 .00 .00 .00* .00 ** .00
F value .45 6.5 *** 3.9 ***
R² .00 0 .54 .86 .62 .85
Note: Dependent variable: Implementation success; N=194.
25
Results: main effects. Examining the results in Table 4 reveals several results. First if the CSFs are examined in Model 2, it can be concluded that not all the 26 factors as identified by Soja (2006) are as important in ERP implementations. Only five factors load with a significant positive value as influencer of successful ERP implementations. The control variable in this study, organizational size, does not have an influence on implementation success. The results show the influence of the factor on the dependent variable in this research, implementation success.
Hypothesis one predicts that the factors in relation to participation are positively associated with implementation success. Co-operation with supplier falls in this category of factors and has a positive influence (β = .16, p < .05) on implementation success, which is consistent with hypothesis one. The empirical findings partially support hypothesis one.
Hypothesis two, stating that top-management participation has a positive influence on successful implementations is not supported. No factors within this category have a positive significant value influencing implementation success.
Hypothesis three, stating that factors related to project definition and organization are positively associated with implementation success is partially supported. Out of the eight factors within this category, two are positively associated with implementation success. Detailed schedule has a positive influence (β = .18, p < .10) on implementation success. Fast effects have a coefficient estimate of .32 and is highly significant (p < .01). The empirical results partly support hypothesis three, in particular detailed schedule and fast effects support hypothesis three. For the other factors in this category, there were no significant findings.
Hypothesis four states that factors in relation to project status are positively associated with successful ERP implementations. The empirical findings do not support this hypothesis. There are no significant positive values for the factors in this category.
26
Figure 3. Results Main Effect on Implementation Success
.18** Implementation participation
(A5) Co-operation with supplier
Top management involvement
Top management involvement
Project definition and organisation (C3) Detailed schedule (C8) Fast effects
Project definition and organisation (C3) Detailed schedule (C8) Fast effects
Project status
Project status
Information system (E1) System reliability (E2) Minimal customization
Information system (E1) System reliability (E2) Minimal customization
Implementation success Implementation success .23*** Control variable Firm size Control variable Firm size .15* .32** .57***
Note: The dashed lines indicate nonsignificant correlations. *Significant at p<.10 (one tailed test);
** Significant at p<.05 (one tailed test); *** Significant at p<.01 (one tailed test).
Results: moderating effects. The further results show the influence of the moderator in this research, namely level of “vanilla” implementation. In Table 4 this is indicated by “Vanilla”.
The data shows that for some of the success factors, the level of “vanilla” implementation has a significant influence. Meaning that for implementations with high levels of “vanilla” implementation, factors are more important than for lower levels of “vanilla” implementation. This goes for the factor motivation system where the relation is positively influenced by high levels of “vanilla” implementation (β = .16, p < .10). This means that for “vanilla” implementations, the positive influence of motivation system on successful implementations is increased and makes it more important. This is in line with hypothesis six and so the empirical findings partially support hypothesis six.
27
customized ERP systems, the influence of detailed scheduling on implementation success is decreased. This is contrary to hypothesis eight and partially rejects it.The influence of the factor pre-implementation analysis on implementation success is negatively influenced by high levels of “vanilla” implementations (β = -.18, p < .10). This factor also falls in the category project definition and organization and partially supports hypothesis eight. Although not hypothesized the reversed effect is also arguable. With low levels of “vanilla”, and so high levels of customization, the effect of pre-implementation analysis on successful implementation gets positively influenced (β = .18, p < .10).
Investment plan is the last factor significantly influenced by implementation strategy. Higher levels of “vanilla” implementations increase the influence of investment plan on successful implementation (β = .13, p < .10). This finding partially supports hypothesis nine.
28
Figure 4. Results Moderation Effects
.15* Implementation participation
(A4) Motivation system
(A5) Co-operation with supplier
Top management involvement (B1) Top management support
Top management involvement
(B1) Top management support
Project definition and organisation (C1) Linking with strategy (C3) Detailed schedule (C4) Pre-implementation analysis (C8) Fast effects
Project definition and organisation
(C1) Linking with strategy
(C3) Detailed schedule (C4) Pre-implementation analysis (C8) Fast effects Project status (D1) Investment plan Project status (D1) Investment plan Information system (E1) System reliability (E2) Minimal customization
Information system
(E1) System reliability
(E2) Minimal customization
Implementation success
Implementation success .22**
.57***
Level of “vanilla” implementation
Level of “vanilla” implementation .20*** -.18* .13* .16* Control variable Firm size Control variable Firm size .00** .15* .28** .22*** .19**
Note: All independent and moderating variables are mean centred; The dashed lines indicate nonsignificant correlations.
*Significant at p<.10 (one tailed test); ** Significant at p<.05 (one tailed test); *** Significant at p<.01 (one tailed test).
Table 5. Summary of Hypothesis Testing Results
Hypothesis Description Resulted Tested in
H1 Critical success factors with relation to participation are positively associated with implementation success. In particular: (a) Project team manager (b) Team composition (c) Team involvement and (d) Cooperation with supplier.
Partially supported
Model 2
H2 Critical success factors with relation to top-management involvement are positively associated with implementation success. In particular: (a) Top-management support, (b) Top-Top-management awareness (c) Top-Top-management participation.
Not supported Model 2
H3 Critical success factors with relation to project definition and organization are positively associated with implementation success. In particular: (a) Linking with strategy, (b) Implementation goals (c) Detailed schedule (d) Pre-implementation analysis (e) Organisational change (f) Monitor and feedback (g) Implementation promotion (h) Fast effect (i) Appropriate training.
Partially supported
Model 2
H4 Critical success factors with relation to project status are positively associated with implementation success. In particular: (a) Investment plan (b) Project team empowerment (c) Financial budget (d) Work time schedule (e) IT infrastructure.
29
H5 Critical success factors with relation to Information system are positivelyassociated with implementation success. In particular: (a) System reliability (b) Legacy system (c) Implementation experience.
Partially supported
Model 2
H6 Higher levels of “vanilla” implementations will increase the effect of implementation participants on implementation success.
Partially supported
Model 3 H7 Higher levels of “vanilla” implementations will increase the effect of top
management involvement on implementation success.
Not supported Model 3 H8 Higher levels of “vanilla” implementations will decrease the effect of project
definition and organization on implementation success
Partially rejected and supported
Model 3
H9 Higher levels of “vanilla” implementations will increase the effect of project status on implementation success
Partially supported
Model 3
H10 Higher levels of “vanilla” implementations will decrease the effect of information systems on implementation success.
Not supported Model 3
Discussion and Conclusion
This research starts from the premise ERP implementations are influenced by different factors. The purpose of this study was to test which factors are important in ERP implementations and more specifically, “vanilla” ERP implementations. The first research question in this study was: “What are the critical success factors for ERP implementations in SMEs?” This research empirically tested the critical success identified by Soja (2006) and discovered a positive significant relationship between five of the factors without taking into account the implementation strategy. (A5) Co-operation with supplier, (C3) detailed schedule, (C8) fast effect, (E1) system reliability and (E2) minimal customization are found to be the necessary elements for project success and success factors for ERP implementations in SMEs. Co-operation with supplier partially confirms hypothesis one. Detailed schedule and Fast effect partially confirm hypothesis three. System reliability and Minimum customization partially confirm hypothesis five.
30
and the implementation is done by those with the know-how. These top managers do not have the technical background and pay less attention to technology (Shiau, Hsu and Wang, 2009).
The second research question in this research was “Which of these critical success factors are more important for a “vanilla” implementation?” For organizations with high levels of “vanilla” implementations, this research has showed that there are specific critical success factors, meaning that these factors increase the possibility for successful implementations with “vanilla” implementations. This is true for three out of the 26 identified by Soja (2006), namely (A4) motivation system, (C3) detailed scheduling and for (D1) investment plan and so these three factors are more important for a “vanilla” implementation.
Motivation system falls within the category of factors for hypothesis six, which states that higher levels of “vanilla” implementations increase the effect of the factors related to implementation participation on implementation success and thus, hypothesis six is partially supported. Although non-significant direct relation is found in this research, motivation system is assumed to be important during “vanilla” implementations since users tend push for technical customization because they want to reduce the amount of change they have to make (Soh and Sia, 2005). Participation in “vanilla” implementations is therefore lower than in technical customization implementations. Motivation systems, with rewarding participation, will increase participation and will therefore be more important during “vanilla” implementations than in technical customization implementations.
Higher levels of “vanilla” implementation also increase the positive effect of detailed scheduling on implementation success. This contradicts hypothesis eight and therefore partially rejects it. Although this rejects hypothesis eight, it is a success factors for a “vanilla” implementation.
31
levels of “vanilla” implementations, there is less complexity, which makes it easier to achieve the planning.The relation of pre-implementation analysis on implementation success is negatively influenced by higher levels of “vanilla” implementation, which is in line with hypothesis eight and therefore partially supports it. Although the direct relation is nonsignificant, pre-implementation analysis is more important for “vanilla” implementations. Business processes need to be altered in “vanilla” implementations based on the system. There is no need to analyse the organization before implementing because these processes will be altered anyway. With higher levels of technical customization, the system will be changed based on the business processes and therefore these need to be analysed. This makes pre-implementation analysis less important for “vanilla” implementations. This result is intriguing because of the negative effect of this implementation strategy on the influence of this factor on implementation success. This means that this factor becomes less important in ERP implementations with higher levels of “vanilla”.
Although not hypothesised, this means that this is a critical success factor for a “comprehensive” implementation strategy. With higher levels of comprehensiveness, the influence of pre-implementation analysis on the successfulness of the pre-implementations is positively magnified.
The influence of investment plan on implementation success is also increased by higher levels of “vanilla” implementations. The direct relation between investment plan is nonsignificant, but this factor can be described as important because it can create a support base for the change which is harder to accomplish in “vanilla” implementations then in customized implementations because users would rather see the system change then that they have to make changes (Soh and Sia, 2005). Investment plan increases the chance for implementation success during “vanilla” implementations and therefore it should be present during this form of implementation. The results of this research will benefit SMEs in implementing ERP systems.
Theoretical Implications
