Requirements Engineering for Enterprise Systems: What We Know and What We Don’t Know? - Intentional Perspectives on Information Systems Engineering - page 124

the cost and effort needed are acceptable. This assumption might not be realistic
in all ES contexts. For example, Soffer et al. [ 54] indicate that it would cost extra
effort to run the OPMmodeling process for aligning a package to enterprise require-
ments, as the analyst first have to create the model of the package functionality. In
our view, it's also interesting to understand how much time (e.g. in person hours),
it would take to create a model of a specific package component, e.g. account-
ing, in a company of a specific size. It is also worthwhile knowing how much
time it would take to learn a modeling/gap analysis technique and its application
process. Answers to these questions are important to make a decision on how to
deploy the technique in a particular context. For example, the first author's per-
sonal experience suggests that a two-day training on the ARIS modeling technique
was not enough for business owners to get comfortable in reading the SAP mod-
els without the help of external consultants. In that case, it turned out that hiring
a specialist in ARIS-modeling on a permanent basis was much more cost-effective
for the ERP-adopter than training all relevant stakeholders on how to use the ARIS
methodology.
Last, the published ES modeling techniques tacitly assume that it's possible to
scale them up to large projects. Today, this type of projects is, more often than not,
cross-organizational, which increases the complexity of ES RE even more. If we are
to apply a process-mining or reverse-engineering based approach to such a setting,
this assumes that all partner companies in a extended enterprise are prepared to
disclose their process and application landscapes (so that the respective tools capture
completely their business environments). Assessing how realistic is to assume this
means including the concept of trust in the discussion on ES requirements. This
alone forms another line of research for the future.
3.3 Did These Techniques Work?
As indicated earlier, many papers on ES elicitation and modeling present empirical
evaluations. In our review, we consistently observed that when authors propose a
technique, they either provide a detailed account of its application in an industrial
setting, or they run a complete action case study research intervention in a com-
pany and reflect on their learning from it. Table 1 presents the type of empirical
research done in the papers which we cited in Sects. 3.1 and 3.2. In this table,
the first column refers to the paper that published a RE-for-ES approach. The
second column reports on whether this approach is for elicitation, or for model-
ing, or for both. The last column indicates the context where the empirical study
in the paper has been done. The table shows that RE-for-ES researchers have
been actively involved in action research with big companies. Some authors also
include empirical research in the IT department of their institution (e.g. studies
on ES implemented in a university). The brief indication of empirical studies in
the table shows that researchers prefer action case studies for their evaluation. This
increases the realism of the study but makes generalizability an important issue to
consider.