An Empirical Taxonomy of Implementation Processes Based on Sequences of Events in Information System Development

A widely accepted and usable taxonomy is a fundamental element in the development of a scientific body of knowledge. However, the creation of good empirical taxonomies of implementation processes is complicated by the need to consider the dynamics of the implementation process. This paper addresses this difficulty by using an optimal matching procedure to measure the pairwise distances among event sequences occurring in 53 computer-based information system (IS) implementation projects. Cluster analysis based on these inter-sequence distances is used to generate the empirical taxonomy of implementation processes. The resulting taxonomy includes six distinct archetypical processes. One of the process types is labeled textbook life cycle (type 4) due to its close resemblance to the detailed, rational approach commonly prescribed in IS textbooks. The logical minimalist process (type 1) follows some of the basic steps of the textbook approach, but is characterized by little project definition and infrequent assignment of personnel. Whereas both textbook life cycle and logical minimalist approaches use external vendors and consultants to some extent, external dependence is much greater in traditional off-the-shelf (type 2) and outsourced cooperative (type 5) processes. The traditional off-the-shelf process simply involves purchasing the system from an external vendor, with little system construction or assignment of personnel. In contrast, the outsourced cooperative process consists of joint system development by internally assigned personnel and external vendors. The remaining two process types—problem-driven minimalist (type 3) and in-house trial and error (type 6)—are both considerably influenced by performance problems. The problem-driven minimalist process is initiated by such problems, with little project definition, and results in a reassignment of organizational roles. The in-house trial-and-error process begins like textbook life cycle, with a clear project definition, but involves frequent system modifications to respond to the performance problems encountered during the project. The paper demonstrates how an empirical taxonomy that incorporates the dynamics of event sequences may be developed. The archetypes comprising the taxonomy are related to other implementation process models available in the literature. Some limitations of the study are acknowledged and its implications for future research and practice are discussed.