Chapter 4.

11

Towards Identifying the Most Important Attributes of ERP Implementations

Piotr Soja

Cracow University of Economics, Poland

Dariusz Put

Cracow University of Economics, Poland

AbstrAct

Enterprise resource planning (ERP) systems have been implemented in various and diverse orga-
nizations. The size of companies, their industry, the environment, and the number of implemented
modules are examples of their heterogeneity. In consequence, a single procedure which leads to the
success of implementation does not appear to exist. Therefore, there have been many implemen-tations
that have failed during, and also after, the implementation process. As a result, a considerable amount
of research has been trying to identify is-sues influencing ultimate project success and also to recognize
the best implementation projects. The aim of this work is to identify the most important characteristics
of ERP implementation which af-fect project success. This study builds on data gath-ered using a
questionnaire directed toward people playing leading roles in ERP implementations in a few dozen
companies. Twelve attributes were identified and divided into three sets representing: effort, effect, and
the synthetic measure of success calculated on the basis of the obtained data. Two agglomeration
methods were employed to identify exemplar and anti-exemplar groups and objects. These elements
were thoroughly analyzed, which led to identifying the most and the least desired attributes of an ERP
implementation project. The findings are discussed and related with the results of prior research. Finally,
implications for practitioners and concluding remarks summarise the chapter.

INtrODUctION

The implementation of an ERP system is a great challenge for a company making the effort of
introducing such a system into its organisation. The implementation project is usually connected with
sizeable expenses for computer software and hardware, as well as for the implementation services
provided by a system solution supplier (e.g., Sarkis & Gunasekaran, 2003). The implementation effects
could be very diverse, beginning from the considerable enhancement of enterprise activity and increase
of its profitability, to the rejection of the system introduced (e.g., Holland et al., 1999; McNurlin &
Sprague, 2002). The companies introducing ERP packages into their organisations differ quite
significantly. The implementation endeavours called ERP projects comprise both simple installations of
single modules of a system and complex solutions dealing with the installation of many system modules
in numerous units of a company (Parr & Shanks, 2000).Therefore, ERP implementation projects form a
very diverse population and in order to compare particular implementations, one has to keep this
diversity in mind so that such a comparison is reasonable (e.g., Stensrud & Myrtveit, 2003). Thus, it
seems appropriate to group purpose-fully implementation projects into homogenous collections, where
the comparison of projects is feasible and sensible. Only in this situation can we talk about a “model”
implementation project and examine the project discovered in order to reveal the most needed
characteristics.Among the methods of projects grouping suggested by prior studies, there are those em-
ploying company size (e.g., Bernroider & Koch, 2001; Buonanno et al., 2005; Everdingen et al., 2000; Loh
& Koh, 2004) and those relying on a criterion of the number of user licenses (Sedera et al., 2003). While
previous research indicates that company size is an important criterion influenc-ing ERP project
conditions, the results regarding the benefits achieved are mixed. Some researchworks suggest that
benefits gained by large and small sized organisations seem to be similar (e.g., Shang & Seddon, 2000;
Soja, 2005) and other studies advocate that benefits differ by company size (Mabert et al., 2003).Prior
studies also suggest other criteria of ERP projects grouping that might influence implementations’
conditions. These criteria include the extent of ERP package modification (Soh & Sia, 2005),
implementation scope and duration time (Soja, 2005, 2006). The results imply that the implementations’
conditions are diverse depend-ing on project type defined by dividing criteria. Moreover, the project
type can have an impact on the effects achieved by a company as a result of ERP implementation. In
particular, the project duration seems to have an important influence on achieved results (Soja,
2005).The multitude of potential factors influencing ERP projects is illustrated by the complex division
presented by Parr and Shanks (2000). They suggest the following categories for the division of projects:
implementation physical scope (single or multiple site), extent of organisational changes, level of system
modification, module implementa-tion strategy, and allocated resources in terms of time and budget.
Taking into consideration the above-mentioned criteria of a division, there are a great many
implementation types. Therefore, Parr and Shanks distinguish three main categories of ERP
implementations: comprehensive, averagely complicated (middle-road) and simple (vanilla).Overall, it
seems that it is hard to find a generally accepted division of ERP projects into groups, which would
constitute homogenous collections of similar implementations. Prior studies suggest various criteria of
ERP projects grouping and these divisions take into consideration merely the variables defining the
efforts made in order to implement a system, but they completely omit the issue of achieved effects.
Meanwhile, incorporating the parameters describing implementation results could lead to interesting
conclusions. The goal of this paper is an attempt to dis-cover the most desired attributes of a model ERP
implementation project. The article is based on research conducted among a few dozen companies
introducing an ERP system into their organisations in Poland. In order to achieve the paper’s goal, the
statistical methods of element grouping were employed, which allowed us to extract the groups of
homogenous projects that were then ordered on the basis of the measure of achieved success. This
procedure allowed us to distinguish the projects having the most desirable characteristics, as well as
those with the least desirable attributes.

LItErAtUrE rEVIEW

AThe idea of discovering issues determining the success of enterprise system (ES) implementation
projects attracted the attention of a considerable number of researchers. There are a great many
research approaches which investigate numer-ous issues and employ various understandings of project
success. The works differ in various aspects, i.e. the employed methodologies and chosen variables.
Some scholars examine the actual implementation process, others focus on the post-implementation
phase, there are also those who treat ES adoption as a continuous endeavour without a clearly defined
end point. The inves-tigated issues vary from technological, through organisational to those connected
with people, and from operational to strategic considerations. The following section summarises major
find-ings achieved by prior research connected with the attributes and characteristics of successful ES
adoption projects.Among many issues analysed by research-ers, knowledge seems to play a paramount
role in enterprise system adoption. In particular, the researchers emphasize the need of adequate
knowledge transfer from external consultants to clients in an adopting organisation (Ko et al., 2005).
This process should be dealt with by sev-eral parties participating in ES adoption: vendors, consultants
and IS/IT managers. They should pay attention to improve not only the quality of ERP products, but also
user’s knowledge and involve-ment. They should also be aware of the importance of choosing suitable
consultants and suppliers (Wu & Wang, 2007). Furthermore, emphasizing the role of knowledge
transfer, McGinnis and Huang (2007) introduced the idea of constituting a knowledge-sharing
community which may play the crucial role of a platform that can be used to provide a common frame
of reference to all ERP activities. The authors also highlight the concept that the actual implementation
does not finish the whole adoption endeavour, which should be still monitored and handled after the
system rollout. The quality of the actual implementation process influences the course of the post-
implementation phase. In particular, Nicolaou (2004) suggests five critical dimensions which affect the
whole adop-tion project, including the post-implementation stage. These issues relate to the review of
overall project scope and planning, the review of driving principles for project development, the
effective-ness of misfit resolution strategies, the evaluation of attained benefits, and the evaluation of
learning.The enterprise system adoption projects differ greatly as regards the scope of the implementa-
tion. In general, the most complicated full-scope and highly-integrated enterprise system adoption
Mprojects are perceived as more likely to bring the best benefits for the company. This issue is
illustrated by Ranganathan and Brown (2006), who conclude that a company’s announcement of ERP
adoption positively influences stock market returns. The authors suggest that the greater the project’s
scope (in terms of number of locations and the extent of introduced functionality) the stronger the
influence.Researchers highlight the need for identifying adequate characteristics of enterprise system in
order to achieve its successful adoption. The enumerated features which appear to be the most
important encompass system quality (Kositanurit et al., 2006) together with other quality dimensions
related to information and service (Chien & Tsaur, 2007), and perceived usefulness of the system
(Amoako-Gyampah, 2007). In particular, on the basis of research conducted among 571 respondents
and employing Technology Accep-tance Model, Amoako-Gyampah (2007) suggests that users’ intention
to use an ERP might depend more on how useful they perceived the system than on their level of
involvement or how easy it will be to use the system. Furthermore, the author advocates that perceived
usefulness influences the users’ involvement and beliefs concerning the necessity of changes. However,
on the other hand, Amoako-Gyampah points out the role of legacy systems and related people’s habits
and expertise, which may negatively influence their attitudes towards ERP.Identifying the desired
features of the enterprise system is a necessary condition for successful adoption. However, equally
important is the ap-propriate implementation process, during which one of the most crucial issues is to
achieve an alignment between the system’s capabilities and the company’s needs. Sharma and Vyas
(2007) emphasize this issue by talking about the synergy between technology and management, which
is advocated as an important element influencing the success of ERP adoption.Enterprise system
adoption is inevitably con-nected with some extent of organisational changes which are carried out in
the adopting company. The more radical approach to organisational changes is called business process
reengineering (BPR), while the less radical method bears the name of business process improvement
(BPI) (Law & Ngai, 2007). Schniederjans and Kim (2003), on the basis of a survey conducted among 115
US electronic manufacturing firms, advocate that business process change should precede the
enterprise system implementation. Furthermore, Law and Ngai (2007), drawing for the experience of 96
companies, conclude that during enterprise system adoption firms should undertake business process
improvement and ought to have dual focus: operational and strategic. The authors demonstrate that
firms which had only opera-tional focus achieved lower performance than those having dual or strategic
focus. In short, the discussion above illustrates that enterprise system adoption should be first and
foremost an organisational and business project. It should be treated as a business-led endeavour, in
contrast to an IT related initiative (e.g., Law & Ngai, 2007; Nicolaou, 2004; Tinham, 2006).The ES
adoption project needs constant monitoring as regards employed resources, both internal and external.
As far as internal resources are considered, Peslak (2006), on the basis of the opinions of over 200 top
financial executives, advocates that cost and time were found to be major determinants of project
success and should be carefully measured and monitored. Further, the author also points out that the
use of external consultants for implementation should be carefully controlled since it was found to
adversely impact cost performance.The enterprise system adoption projects form a very diverse group
differing in several aspects. This issue is discussed by Stensrud and Myrtveit (2003), who examined 30
ERP projects using Data Envelopment Analysis Variable Returns to Scale (DEA VRS) to measure the
productivity of software projects and employed the method to identify outstanding projects of ERP
systems that may serve as role models. As a result, they suggest that the average efficiency among
investigated projects is approximately 50%. Furthermore, the authors notice that there were significant
differences in productivity between projects in various industries. Therefore, one should exhibit caution
when benchmarking and comparing projects across industries. The authors suggest that performance
assessments should include both productivity and quality indicators, and it should also take into account
other external factors such as schedule constraints.

As regards the difficulties with the comparison of different ES adoption projects, the impact of
organisational size on ES project conditions was studied by several researchers. The majority of authors
defined organisational size taking into consideration the number of employees; other studies
understood the size of a company in terms of the level of revenues (Mabert et al., 2003) or defined the
size of an ES project as a number of installed licences of the system (Sedera et al., 2003). While
investigating ES projects, the schol-ars employed research approaches based on case studies, interviews
and surveys. Their respondents were mainly adopters; however, some studies also enquired system
supplier representatives (e.g., Mabert et al., 2003).

The results of prior works illustrate that in the case of small firms, the most important issues comprise
available human resources and system fit into company’s organisation (Bernroider & Koch, 2001;
Buonanno et al., 2005; Mabert et al., 2003; Muscatello et al., 2003; Raymond & Uwizeyemungu, 2007),
which result in a shorter implementation time, lower costs, and lack of the need for significant
organisational changes (Adam & O’Doherty, 2000; Bernroider & Koch, 2001; Mabert et al., 2003). The
paramount significance of human resources in the case of small and medium-size companies is
expressed by Sun et al. (2005), who claim that a great emphasis during ERP implementation should be
placed on people.

It is worth noting that the issues connected with organisational changes are not perceived
unambiguously by researchers: some point out the lack of willingness to perform organisational changes
in the case of small firms, others claim that small companies are more likely to change their processes to
fit the system (Mabert et al., 2003). The cited research works also suggest that benefits realized by small
and large companies differ: larger firms achieve greater benefits, with a special emphasis on a financial
indicators improve-ment, while smaller companies accomplish first and foremost the improvement of
manufacturing and logistics activities. The idea of limited benefits achieved by small and medium-size
companies is reflected in the research of Sun et al. (2005), who conclude that as the implementation
schedule increases, the cost increases accordingly, while the achievement increases to some point,
beyond which there is no significant achievement benefit.

rEsEArcH MEtHODOLOGy

This study is based on exploratory research conducted among enterprises introducing an ERP package
into their organisations. A field study was adopted as a general research approach, and a questionnaire
was employed as a data-gathering technique (e.g., Boudreau et al., 2001). The research question posed
in this study could be expressed as follows: What are the most desired attributes of an ERP
implementation project?

The research questionnaire was comprised of questions with a mixture of scale, multiple choice and
open questions. The purpose of these queries was to provide demographic data and details necessary to
assess project conditions and implementation effects. The list of respondent enterprises was prepared
on the basis of reports analysing the ERP market in Poland and databases containing companies’ address
data. The resulting list contains firms that introduced an ERP package into their organisations with a
broad scope, estimated on the basis of available data.

The questionnaire was directed toward the people playing leading roles (the project leader, if possible)
in the implementation. With the help of the questionnaire, data has been gathered regarding the
conditions of implementation projects, efforts incurred, as well as results achieved. The collected data
contain various pieces of information regard-ing both the implementation process and achieved results.
Part of the data contains objective items, while the other (subjective) include respondents’ individual
evaluations. The achieved collection of projects is varied; hence, an attempt to identify the model
object, having the most desired attributes which led to the completion of implementation goals, is not
an easy task.

The group of objects was characterised by 12 attributes, which were divided into 3 distinct subsets. In
the first subset, there were input indi-cators of an implementation process—let us call them “effort
indicators”. The second group was comprised of variables relating to the implementa-tion results—
called “effect indicators” (see Table 1). The third, one-element subset, contained the calculated
variable being a synthetic measure of implementation success, which was calculated on the basis of data
gathered from the enterprises (Soja, 2004). In the next stage of the research, this measure was used to
establish the hierarchy of the groups.

The success synthetic measure, based on the understanding of success in the information sys-tems
domain (e.g., Lyytinen, 1988), employs 5 partial measures: (1) the actual scope of an imple-mentation
with respect to the planned implementation, (2) the actual duration with respect to the assumed
duration, (3) financial budget with regard to the planned budget, (4) users ’ level of satisfac-tion from
the system introduced, and (5) the ex-istence and achievement of project goals (Soja, 2006).

In the research conducted, two agglomeration methods were employed: hierarchical Ward’s method
(e.g., El-Hamdouchi & Willett, 1986), which is the most commonly used agglomerative method
employed for forming clusters (Everitt, 1993) and non-hierarchical k-Means method. The aim of
Ward’s method is to join objects together into ever increasing sizes of clusters using a measure of
similarity of distance. K-means, on the other hand, is a simple non-parametric clus-tering method that
minimises the within-cluster variability and maximises the between cluster variability. The k-means
method requires that the number of clusters is specified beforehand (National Statistics 2001 area
classification, 2001). One possibility for obtaining the number is to run Ward’s method and use the
outcome as initial configuration for k-means.

Since in the k-Means method a researcher has to arbitrarily provide the number of clusters, the two-
phased approach is common in cluster analysis research. In the first stage, the hierarchical method is
applied in order to determine a preliminary number of clusters (e.g., Ward’s method), and in the
second step the actual classification of objects using the k-Means method takes place (e.g., Everitt et al.,
2001). This approach was adopted during this research and is illustrated in Figure 1.

The procedure is aimed at the separation of object groups which are similar to each other but differing
to a greater extent from the objects be-longing to the remaining groups (e.g., Kaufman & Rousseeuw,
1990). Firstly, the standardization of variables was carried out, which allowed us to remove the
excessive influence of variables having a wide range of values from the outcome of the research. In the
next step, the Ward’s hierar-chical grouping method was applied. On the basis of a distance diagram
obtained, two decisions were made:1. Some objects were excluded from further processing. The objects
most dissimilar to other items or those forming small, two element groups, were treated as accidental
measurement. Their exclusion allowed us, at the next stage, to receive more homogenous clusters,
containing objects the most similar to each other and laying closer to the hypothetical centre of a
cluster.

2. The k value was selected for the applied k-Means method. The greater the k, the more the clusters.
These clusters tend to be smaller and contain more similar objects. A small kmeans, on the other hand,
fewer groups and more diverse objects within each subset. In order to determine the k value, the
distance diagram achieved with the use of Ward’s method was employed. After excluding objects
dissimilar to other items, the visual analysis of number clusters in the diagram was performed.

During the next stage, the k-Means method was applied. The calculations were performed three times:
(1) for effort indicators only, (2) for effect indicators only, and (3) for all eleven in-dicators of effort and
effect together. As a result, the separated groups of similar objects were ex-tracted, together with the
distance of each object from the hypothetical centre of a cluster. For each cluster, the average value of
success measure was calculated using synthetic success measures evaluated for objects belonging to a
particular group, and on the basis of this value, hierarchy of the group was determined. The cluster
having the greatest average value of success measures was recognised as containing objects with the
most desired characteristics. Simultaneously, the cluster with the least average value of success
measures was recognised as having objects with the least desirable attributes.

Within each of these two extreme groups, one object having the smallest distance from the hypothetical
centre of a cluster was distinguished. The object coming from “the best” group was regarded as
exemplar (a model implementation), while the object extracted from the group having the least average
value of success measure was perceived as anti-exemplar (an anti-model implementation). Since the
calculation was performed three times, three exemplars and three anti-exemplars were extracted,
characterising the most needed and, also, the least desirable attributes of variables describing efforts,
effects as well as efforts and effects jointly (in some cases two objects were distinguished, since both
were equidistant from the centre).

The detailed analysis of the attributes of exemplar and anti-exemplar objects, as well as basic statistics
calculated for clusters distinguished as the best and the worst, allowed us to draw conclu-sions as
regards to the most needed and the least desired parameters of a implementation project. Thus, the
research question can be answered, i.e. the most desired attributed of an ERP project can be elicited.
These conclusions could be a sug-gestion for people responsible for running ERP implementation
projects, so that they pay atten-tion to certain facts, which contribute to project success or failure.

rEsEArcH DAtA

During the research, 223 enterprises were contacted and 68 (30%) answers were obtained from
enterprises representing the whole country and various industries. From among the questionnaires
received, 64 were accepted for further analysis. All enterprises investigated in this study represent
companies which introduced an ERP system into their organisations. The companies classified by
industry type are described as shown in Table 2, where the number of companies belonging to particular
industries was provided. As can be easily seen, the vast majority of companies comprise of
manufacturing enterprises (75%).

For the purpose of analysis, this study ad-opted the criterion defining enterprise size as the number of
employees. The understanding of “small” and “large” companies is derived from the European
Community’s definition for small and medium-sized companies (e.g., The Commis-sion of the
European Community, 1996).
The investigated enterprises differ significantly in their size regarding the number of employees, which
can be seen in Table 3. It contains, in sub-sequent rows, the number of companies (columnn) employing
a number of workers which falls within a specified range. The largest group is formed by companies
employing from 501 to 1000 workers, and constitutes more than one fifth of the companies researched.
The second larg-est group is made up by the biggest companies employing over 1000 workers, which
represents 20% of enterprises evaluated. Certainly, the least numerous group is formed by small
companies, employing not more than 100 workers.

The implementation projects researched make up quite a diverse group when project duration time is
taken into consideration. Among the companies examined, there are projects lasting not more than a
couple of months, as well as implementations with a duration time longer than 3 years. Table 4
illustrates the number of projects as regards planned and actual duration time.The examined projects
are also diverse as regards to the implementation scope defined by