Information technology (IT) has been identified as a major organizational resource that enables organizations to achieve competitive advantage or to survive in their market (Brynjolfsson & Hitt, 1996). As business and IT should nowadays cooperate seamlessly, the rigidity of old IT proved to be a bottleneck for sustaining competitive advantage in the markets.

On their journey to overcome those rigid IT structures, and instead create flexible and cost-effective information landscapes, organizations identify a need for Service Oriented Architecture (SOA). SOA can be seen as the next episode in the maturity of IT as a professional discipline and as a next step in the way business can make use of IT (Van den Berg et al., 2007, p. 17). SOA is an architectural style in which systems are divided into (loosely coupled) autonomous services that, when combined, guarantee extremely flexible and controllable IT environments. By defining services at elementary functionality level, those services function as stable ‘building blocks’, and can be used to quickly realize new products and services. Because those services have potential of reuse, SOA has the promise to reduce maintenance and management costs. Because of agility and flexibility, SOA can have significant impact on the business capabilities of an organization (Krafzig et al., 2005).

There are few academic empirical studies on SOA, and more specific there is no research conducted to factors that influence the implementation of this technology. Given that business shows high interest in implementing SOA (Forrester, 2006; Gartner, 2007), but at the same time identify the risk of failed SOA projects (IDC, 2006) there is a true need to answer the question “what contextual factors are critical for the successful introduction of a Service oriented architecture in an organization?”.

In three steps the initial research model was built. First, a frame of reference was defined with 6 technologies, each satisfying the criteria that (i) they can be qualified as infrastructure IT project and (ii) that they are described in a mature body of literature. Second, relevant factor literature of those technologies is used to derive at a set of factors that are assumed to be important. Third, a quick scan of face validity and completeness has been conducted by consulting IBM experts and 10 respondents from industry (Enterprise Architect level). This effort resulted in a relevant research model consisting of twelve factors, what resulted in twelve hypotheses representing the relation between the factor and overall perceived SOA implementation success.

The research model served as input for the discussions with eleven respondents from industry (nine companies across five industries). Using a dual inquiry mode (adapted Delphi method) respondents first completed an online survey and scored the importance of factors for their specific case, and provided contextual information about their organization and the project’s success. Based on this first round of inquiry we planned follow-up interviews at those companies, and elaborated on their initial answers, which were compared to the answers of other organizations.

The study shows that eleven of the factors proved relevant, one factor (user participation) was excluded, and a new factor was included (relationship with SOA partner). Also, we found that respondents clearly differentiate between business- and IT-driven SOA projects, and the critical factors differ for both types of projects. In retrospect we found support in literature for this differentiation. In figure 1 we present the resulting set of hypotheses. Respondents argue that IT- driven implementation factors are critical for the success of all projects, while projects with a specific business focus also experience the importance of a set of additional factors. Next to indicating what factors are important, we reported how the factors should successfully get managed, and what benefits the respective factor brings for the implementation process.

Next to the new and validated set of hypotheses, interesting conclusions can be drawn. Generic IT implementation factors proved to be important for SOA projects. Hence, we conclude that our fundamental assumption that SOA can be characterized as an IT system is supported. Next, specific factors are important for SOA. We found that SOA, in relation to other information systems in our framework of reference, is more seen as a ‘back-end’ solution for both IT and business change, with a larger project scope than found in other systems. Hence, user interaction is less important while factors keeping track of the large project and organizational context increase in their relative importance.

As presented we found that respondents differentiate between the factors critical for business- and IT- driven implementations. We argue that implementations can have different drivers on the short term. However, on the long term we see all implementations urging for business change, as business becomes interested in exploiting the full benefits of SOA. Having neglected business-related implementation factors in the beginning of each project, urges for high workaround costs at later stages to get business needs integrated in the project.

More research is required to further develop our understanding of SOA implementations. The presented set of validated hypotheses offers interesting directions for further quantitative of qualitative research.

Share / deel