EMPIRICAL STUDY OF ERP SYSTEMS IMPLEMENTATION
COSTS IN SWISS SMES
Catherine Equey
1
, Rob J. Kusters
2
, Sacha Varone
1
and Nicolas Montandon
1
1
Haute Ecole de Gestion, 1227 Carouge, Switzerland;
2
University of Technology and Open University, The Netherlands
Keywords: ERP implementation; cost drivers; Swiss SMEs.
Abstract: Based on the sparse literature investigating the cost of ERP systems implementation, our research uses data
from a survey of Swiss SMEs having implemented ERP in order to test cost drivers. The main innovation is
the proposition of an additional classification of cost drivers that focuses on the enterprise itself, rather than
on ERP. Particular attention is given to consulting fees as a major factor of implementation cost and a new
major cost driver has come to light. “Consultant experience”, not previously mentioned as such in literature,
appears as an important aspect of ERP implementation cost. Particular attention must be paid to this factor
by the ERP implementation project manager.
1 INTRODUCTION
One of the main questions asked by management in
charge of an Enterprise Resource Planning (ERP)
project is “How much does it cost”? It is very
difficult to provide a direct answer to this question
and academic literature prefers to investigate cost
drivers (Kusters, Heemstra and Jonker, 2007); while
remaining otherwise scarce on the matter of the level
of cost. The aim of this research is to examine
factors of implementation cost and specifically the
research question “which factors substantially
influence ERP implementation cost?”
The issue of factors that impact ERP
implementation cost is discussed explicitly by
Stensrud (2001). In his research, he wondered if the
existing body of knowledge developed for software
cost estimation was applicable to estimation of ERP
implementation effort.
Several software cost estimation approaches
exist, such as the constructive cost model
(COCOMO) developed by Boehm (1983). The
approach states that under normal circumstances
development costs are a function of project size.
Since the circumstances in which a project takes
place are rarely ‘normal’, the estimate must be
refined using additional cost drivers. As an example,
one of the models proposed by Boehm is as follows:
Development costs =(a *cd[size]
b
)
* cd
1
* cd
2
…* cd
14
(where cd means cost driver)
(1)
The cost driver ‘size’ (cd[size]) is viewed as the
most dominant cost driver, not only in COCOMO
but also in many other models (Kusters,
Van Genuchten and Heemstra, 1990).
Stensrud (2001) concluded that since most
software cost estimation (SCE) approaches are based
upon the use of the number of lines of source code
(Boehm, 1983) or some synthetic variable such as
function points (Albrecht and Gaffney, 1983) to
assess the size of the project, these approaches are
not immediately applicable. An ERP implementation
project may contain some software development, but
will also contain substantial modelling, installation
and reorganization efforts. It seems unlikely that a
one-dimensional measure of software size will
capture this complexity. He did however conclude,
that a measure of size for an ERP implementation
project would likely be multi-dimensional; using a
combination of measures such as the number of
users, the number of reports that have to be
designed, and the number of ERP modules.
Stensrud (2001) further concludes, based on a
screening of existing SCE tools, that the concepts
provided by parametric models such as COCOMO II
(“COCOMO II”, 1998) provide a good starting point
for the development of an estimation model. Crucial
elements in these models are the existence of a size
metric that can be used to estimate ‘normal’ costs, as
well as cost drivers that can adjust for project
specific issues. He also concluded that emergent
143
Equey C., J. Kusters R., Varone S. and Montandon N. (2008).
EMPIRICAL STUDY OF ERP SYSTEMS IMPLEMENTATION COSTS IN SWISS SMES.
In Proceedings of the Tenth International Conference on Enterprise Information Systems - DISI, pages 143-148
DOI: 10.5220/0001683601430148
Copyright
c
 SciTePress

models for estimation of implementation effort of
standard software, in particular the COCOTS model
(“COCOTS”, 2001), may provide valuable support
in this area. Empirical works by Francalanci (2001),
Von Arb (2001), and Kusters et al. (2007) support
this line of reasoning.
The research by Francalanci (2001) is focused on
the identification of a usable measure of size. In
agreement with Stensrud (2001), she deduces that
such a measure should be multi-dimensional. Based
on data from 43 SAP R/3 implementation projects in
a number of European companies, she identifies
three constituting elements for such a size metric:
• Size of the organization: The size of an
organization reflects its inertia, its ability to resist
change. The assumption is that the larger and
more cumbersome an organization, and in
consequence the more inert it is, the more an
implementation effort will cost. As measures of
organizational size she tested the number of
employees and revenue. Both were found to be
useful.
• Size of the configuration: The size of the
configuration effort is expressed in the number
of modules or sub-modules that are to be
implemented. The assumption is that effort will
increase with the number of modules to be
implemented.
• Size of the implementation: Implementation
effort is expressed by the number of users
involved, since these indicate training and
reorganization effort.
Like Francalanci (2001), Von Arb (1997) focuses
solely on size. In his research, a multi-dimensional
measure based on number of users and number of
(sub-) modules is identified. These results are fairly
similar to Francalanci’s, but do not look at
organizational size.
Kusters et al. (2007) looked at both size and at
additional cost drivers in an in depth investigation
into two companies. The notion that size is
multidimensional was supported in both
organisations, but the composing metrics were
different. It appeared, as exposed in Table 1, that
size was perceived as a combination of:
a) A measure related to the amount of work that is
involved in configuring the ERP system. For this
measure, items such as the number and
complexity of transactions, interfaces, reports
and the amount of data and data conversion were
mentioned. In practice, people perceive “size”
related to the configuration effort at a more
detailed level (e.g. number of interfaces) than do
both Francalanci and Von Arb, who look at the
rather coarse measure number of modules.
b) A measure indicating system implementation and
business reorganisation costs. Francalanci (2001)
refers to this as implementation size. As this
“size” increases, more staff needs to be trained
and also more people are involved in
organisational change efforts. This measure
includes items such as number of users, number
of user groups.
c) Francalanci’s ‘size of the organisation’ was also
referred to explicitly, but notions of number of
user groups’ or number of departments could be
construed as such.
It is unclear how to consider the cost driver number
and complexity of business processes mentioned by
Organization II. Complex business processes almost
certainly have an impact on the configuration effort.
Process modelling is a standard part of ERP
implementation preparation and a large number of
process model metrics are already available (see for
example Netjes, Limam Mansar, Reijers, and Van
der Aalst, 2007).
Table 1: Results on size related cost drivers.
There seems to be consensus in available literature
on the usefulness of a multi-dimensional size related
cost driver. There is also some agreement as to the
dimensions involved, but definitely more research is
required into the metrics to be used for each
dimension.
The ‘configuration effort’ dimension is
mentioned by all three references. Francalanci
(2001) and Von Arb (1997) used number of (sub)
modules as a metric, but Kusters et al (2007)
rejected this notion and proposed more detailed
metrics.
Francalanci; Kusters et al:
Von Arb Organization I Organization II
- № of (sub)
modules
- size of
organization
- № of users
- № of
transactions,
- № of
interfaces,
- № of reports
- amount data
conversion
- № of user
groups
- № of users
- № and complexity
of transactions
- № and complexity
of interfaces
- № and complexity
of reports
- size and
complexity of data
- № of departments
- № of users
- № and complexity
of business
processes
ICEIS 2008 - International Conference on Enterprise Information Systems
144

‘Implementation effort’ is also an accepted
dimension. The metric most mentioned is number of
users. However, not just training and motivation
effort are important. The degree of reorganisation
required can be expected to play a role. This is
confirmed by Kusters et al (2007), where cost
drivers such as fit between organization and
product, process maturity and insight in the
processes were mentioned as additional cost drivers.
This leads to a test of an additional metric: degree of
BPR.
Organisational size or ‘planning effort’ is the
third dimension that was previously mentioned.
Apart from the size related factors discussed
above, people related factors are most likely to
impact ERP implementation costs (see Boehm,
1983, and Kusters et al, 1990, for general arguments
to this effect; and Kusters et al, 2007, for ERP
specific results).
Given the availability of data from a study of
Swiss SMEs in 2006 (Equey, 2006), it is interesting
to take a closer look at the dimensions involved. The
aim is to substantiate these three efforts on the basis
of this specific population. As far as we know, this is
the only existing empirical study based on a broad
based survey of Swiss SMEs.
This section focuses on existing literature. In
Section 2, we present the sampling strategy of our
survey of Swiss SMEs. Section 3 presents
descriptive statistics of our data followed by a
correlation analysis. In conclusion, we point out the
main findings, the limitations of this study as well as
directions for future research.
2 METHODOLOGY
The statistical evidence for this study was collected
on the basis of a written survey. The first phase of
the research consisted of in-depth interviews of
Swiss companies from the French speaking part of
the country. This multiple case study (Equey & Rey,
2004) revealed a number of research questions and
associated hypotheses that lead to the design of the
questionnaire. The questionnaire was conceived with
the participation of senior consultants from the four
major vendors of ERP solutions for SMEs on the
Swiss market (Abacus, Microsoft, Oracle and SAP).
The final version of the survey was broken down as
follows: contact details, activities and financial
information about the company, specificities of
implemented ERP, description of the
implementation process, project organization,
outcome and benefits derived from the use of the
ERP system, difficulties and problems encountered.
The main purpose of the survey was to determine
the extent to which Swiss SMEs were aware of or
have implemented ERP. The questionnaire covered a
wide range of topics including implementation and
organisational factors but also issues such as user
satisfaction, the tools used and the perceived value-
added. In the present paper we focus only on the
data pertaining to costs.
More than 4’000 Swiss SMEs were contacted
over a six-month period between November 2005
and April 2006 to take part in the nation-wide
survey. The questionnaire was written in French,
German, Italian and English and was distributed by
post. An electronic version was also made available.
The French version is included in (Equey, 2006).
Other versions are available from the authors.
Contact details for SMEs were obtained from the
Swiss federal office of statistics (OFS) and a sample
was constructed according to the following three
criteria: the size of the company in terms of the
number of employees, the sector of activity
(secondary/ tertiary) and the linguistic region.
In order to be demographically representative,
75% of the sample was chosen from the German-
speaking region of Switzerland, 20% from the
French-speaking region and the remaining 5% from
the Italian-speaking region. In addition, 84% of the
companies surveyed employed from 1 to 49 persons
and the remaining 16% employed between 50 and
249 persons.
To obtain the relatively high response rate of
17.2%, the mailing was followed up by a telephone
interview. A total of 687 Swiss SMEs responded to
the questionnaire. Of those, 18.2% indicated the use
of an ERP, whereas 81.5% or 560 declared not using
an ERP. These results show a relatively low level of
penetration in Swiss SMEs (less than 20%).
3 RESULTS AND DISCUSSION
3.1 Descriptive Statistics
This paper uses the data from the survey in Equey
(2006) and in particular, the sub set of 125
respondent ERP users who had completed the
detailed questionnaire. The inquiry revealed certain
trends that are summarized below.
The respondents indicated a project timeframe of
less than one year in 80% of cases and even less than
six months for 53%. These projects generally
involve less than 7% of the company’s internal staff
EMPIRICAL STUDY OF ERP SYSTEMS IMPLEMENTATION COSTS IN SWISS SMES
145

and financially represent no more than 1% of annual
revenue in 35% of cases (cf. Table 2). A further 38%
fall within 1% and 3% of annual revenue. On
average, 4 modules are implemented in these
projects. Unsurprisingly, the finance module is used
in over 80% of cases and the other most frequently
utilized modules: Purchasing, HR, Inventory
management and CRM appear in over 50% of
responses. On the other hand, the production
module is used by fewer than 40% of respondents,
highlighting the preponderance of tertiary sector
enterprises in our sample.
The data reveals in most cases a ratio of one
(external) consultant employed to each staff member
committed to the project. The cost of consulting was
under 20% of the total project cost in 57% of cases
with a further 20% falling within 50% of total cost
(cf. Table 4b). Consulting cost is clearly the main
individual factor of total cost of implementation in
ERP projects.
Software user licenses represent roughly 15% of
total project cost in 54% of cases (cf. Table 4c) and
the ongoing commitment to maintenance is on
average less than 0.5% of the company’s annual
turnover (cf. Table 3). Half of the companies
revealed that the number of end users of the system
was less than 10 and a further 44% had between 10
and 100 end users.
The overall costs of the projects covered by this
survey (Equey, 2006) are shown in Table 2. The
results are somewhat surprising since
implementation and maintenance costs were
expected to be higher. Table 3 lays out the average
ongoing costs of the systems implemented and
Tables 4a, 4b and 4c show a breakdown of the
implementation costs.
Table 2: Total cost of implementation of an ERP.
Total cost in % of revenue % of companies
Less than 1% 35%
Between 1 and 3% 38%
Between 3 and 5% 14 %
Between 5 and 7% 2%
Greater than 7% 2%
Did not respond 9%
Table 3: Ongoing costs of an ERP.
% of annual revenue Outsourcing
% in category
Maintenance
% in category
Less than 0.5% 67% 64%
Between 0.5 and 1% 7% 21%
Greater than 1% 5% 5%
Did not respond 21% 10%
Table 4a: Investment in IT during ERP implementation.
% of total cost % of respondents
Less than 5% 28%
Between 6 and 10% 20%
Between 10 and 20% 26%
Greater than 20% 17%
Did not respond 19%
Table 4b: Consulting costs as a percent of total costs.
% of total cost % in each category
Less than 20% 57%
Between 20 and 50% 20%
Between 50 and 70% 8%
Greater than 70% 6%
Did not respond 19%
Table 4c: User licenses as a percent of total cost.
% of total cost % in each category
Less than 10% 34%
Between 10 and 15% 20%
Between 15 and 20% 11%
Greater than 20% 24%
Did not respond 11%
These findings are interesting in their own right
but give no information about the cost drivers
involved.
3.2 Correlation Analysis
In this research, we mainly look at factors that
substantially influence ERP implementation costs.
Some of the variables poorly fit existing
classifications but are significant in terms of their
impact; one example is Consultant’s level of
experience. We therefore propose an additional
classification that seems more appropriate within the
context of our variables.
The variables are classified into three groups:
enterprise characteristics, people and
implementation. This classification focuses on the
enterprise, rather than on the ERP itself, as
previously proposed by Kusters et al (2007) and
Francalanci (2001), to assist in the decision making
process of enterprises for its cost calculations.
To ascertain if a relationship exists between the
cost of an ERP project and a variable, we use a
measure of correlation r, which indicates if a linear
relationship exists between variables. We also
calculate the probability (the p-value) that such
relationship exists by chance only. As a standard,
relationships having a p-value of less than 5% are
ICEIS 2008 - International Conference on Enterprise Information Systems
146

deemed to be present by chance only and are thus
rejected.
Enterprise Characteristics. The cost of an ERP
project has been found to be dependent upon annual
sales revenue r = -0.167, p (one-tailed) = 0.44 as
well as the fact that the enterprise is a subsidiary of a
foreign holding r = -0.244, p = 0.01. The negative
correlation indicates that the increase/decrease of
one variable corresponds to the opposite for the
other variable. This is normal since the cost has been
coded as a percentage of the annual sales and
therefore, once a limit has been reached in the
amount of money for an ERP system, the percentage
of annual sales for the cost decreases. There is also
a strong correlation between annual sales and the
fact that the enterprise is a subsidiary of a foreign
holding r = -0.321, p = 0.01. Therefore, the cost of
ERP is found to be lower for these subsidiaries.
Those two variables can be considered as a (single)
factor for which a relationship exists with the cost.
No evidence of a relationship has been detected
between the cost of an ERP project and the number
of ERP users, the number of employees or the sector
of activity. Prevailing assumptions about “ERP
users” cannot be validated with our data.
People. The cost of an ERP project has been found
to be dependent upon the management's involvement
r = 0.182, p = 0.029; the ERP consultant's level of
experience r = -0.241 p = 0.006; the employee's
involvement r = 0.171 p = 0.033; the ratio of
external consultant by internal employee r = 0.172,
p = 0.038. It is interesting to note that even if the
cost of an experienced consultant is high; his
experience will probably decrease the duration of
ERP implementation and subsequently the total cost
of the project. The involvement of employees in the
ERP project increases the cost, but this involvement
may probably be considered as a way to facilitate the
adoption of a new system.
No evidence of a relationship has been detected
between the cost of an ERP project and the
employees' qualifications or field of expertise, nor
with the project manager's position.
Implementation. The cost of an ERP project is
found to be dependent upon the number of modules
to be implemented r = 0.186, p = 0.022; Pearson’s
correlation coefficient indicates a positive
relationship between those variables. This justifies,
in support of literature, the a priori intuition that as
the number of implemented modules increases, so
does the cost of an ERP project. There is a less than
3 percent chance that a correlation coefficient this
large would exist by chance only.
The cost of an ERP project is also dependent
upon the type of module(s) implemented. Indeed,
some of the modules are found to be positively
correlated to the cost of an ERP project, since their
significance values are no more than 0.05. There is a
medium intensity of the relation between those
modules and the cost of an ERP project
(cf. Table 5).
Table 5: Correlation to cost of ERP project.
ERP modules are shown to be related with a factor
analysis. Two main factors have been detected: a
first one includes the procurement (SCM) module,
the production module, the sales /CRM module and
the inventory module; a second one includes the
finance module and the human resource module.
The relationship between modules can be
summarized as follows, where only significant
correlations are shown.
Figure 1: Correlation between modules.
On the other hand, the results show no strong
evidence of a relationship between the cost of an
ERP project and the other types of modules
individually (i.e. finance, human resources, decision
making, project management...). Moreover, no
evidence of a relationship has been found with
organization tool used or the ERP architecture (web
server or client).
Module r p
Procurement (SCM) 0.260 0.005
Production 0.220 0.017
Sales / CRM 0.274 0.003
Inventory 0.186 0.045
EMPIRICAL STUDY OF ERP SYSTEMS IMPLEMENTATION COSTS IN SWISS SMES
147

4 CONCLUSIONS
This research paper points to some of the factors that
may influence the cost of an ERP project. An
additional classification of these cost drivers has
been introduced that is focused on the enterprise,
rather than on ERP.
An important cost driver mentioned in the
literature is clearly validated by our analysis: the
cost of on ERP project is dependent on the number
of modules to be implemented. On the other hand,
the interdependence of the number of ERP users and
the ERP project cost could not be established with
enough reliability through our data. A usual belief is
that the cost of the user licenses is a central factor of
cost. Nevertheless, our analysis has not revealed
evidence of such a relationship. Project managers
often, counter productively, “over focus” here to try
to generate savings.
The importance of the factor consulting cost
clearly stands out in the data and our analysis reveals
a new major cost driver, not discussed as such in
literature, relating to consultant experience. That
people characteristics would impact project costs is
not a surprise in itself. However, that this impact is
so important, that it ranks with size and can be
identified by such a correlation analysis is certainly
surprising. Consulting is implicit to other cost
drivers such as the number of interfaces or reports
and thus deserves investigation. It is interesting to
point out the negative correlation found between
consultants’ experience in ERP and total cost. This
result implies managerial and practical consequences
concerning the choice of consultants.
As empirical research attempts to measure
business perceptions, some limitations or biases are
unavoidable. Consequently, as is always the case in
empirical research, results should be interpreted with
some caution. The extrapolation of these results to
large companies is not appropriate, and future
research should, therefore, be conducted for them.
Further research to construct a multiple regression
model will be a next step, in order that managers
may evaluate ERP implementation project cost
based on enterprise characteristics.
REFERENCES
Albrecht, A.J., & Gaffney, J.E. (1983). Software Function,
Source Lines of Code, and Development Effort
Prediction, IEEE Tr. on Software Engineering, SE-
9(6).
Arb, R. von (1997). Vorgehensweisen und Erfahrungen
bei der Einführung von Enterprise-Management-
Systemen, PhD. Dissertation Universität Bern,
Switzerland.
Boehm, B.W. (1983). Software Engineering Economics,
Prentice Hall.
COCOMO II, Model Definition Manual (1998).
http://sunset.usc.edu/csse/research/COCOMOII/cocom
o_main.html (Retrieved 20.11.2007).
COCOTS Model Description, (2001).
http://sunset.usc.edu/research/COCOTS/index.html
(Retrieved 21.11.2007).
Equey, C., & Rey, A. (2004). La mise en place d’une
solution de gestion moderne (ERP/PGI), quels enjeux
pour une PME/PMI ? 1ère partie : étude de cas
détaillés, Working paper N°HES-SO/HEG-GE/C--
06/1/4--CH
Equey, C. (2006). Étude du comportement des PME/PMI
suisses en matière d’adoption de système de gestion
intégré, Working paper N°HES-SO/HEG-GE/C--
06/12/1--CH
Francalanci, C. (2001). Predicting the Implementation
Effort on ERP projects, Journal of Information
Technology, 16(1), 33-48.
Kusters, R.J., HeemstraF.J., & Jonker, A. (2007).
Determining the costs of ERP implementation,
Proceedings of the 9
th
International Conference on
Enterprise Information Systems, Vol. Database and
Information Systems Integration, 102-110.
Kusters, R.J., Van Genuchten, M., & Heemstra, F.J.
(1990). Are software cost estimation models accurate?
Information and Software technology, 32 (3), 187-190.
Netjes, M., Limam Mansar, S., Reijers, H.A., & Aalst,
W.M.P. van der (2007). An evolutionary approach for
business process design: towards an intelligent system,
Proceedings of the 9
th
International Conference on
Enterprise Information Systems, Vol. Information
Systems Analysis and Specification, 47-54.
Stensrud, E. (2001). Alternative Approaches to Effort
Prediction of ERP Projects, Information and Software
Technology, 43, 413-423.
ICEIS 2008 - International Conference on Enterprise Information Systems
148