Empirically Driven Improvement of Generic Process Models

We are involved in an experiment to design an improvement process for a generic software process model (henceforth generic model) in collaboration with our industrial partner. Our partner's main line of business is the marketing and evolution of a generic model targeted at the development and maintenance of business information systems. Currently, the generic model has been licensed to over 100 organisations, some of whom are also taking part in the experiment. The original version of the generic model was based on the experiences of key personnel in developing large information systems. However, as the clientele base grew it was realised that the generic model was being applied to projects and organisations very di erent from those of the original experiences. This resulted in greater dependence on the skills of the company's personnel to modify the generic model as required in the eld. Furthermore, the existing generic model improvement process is operating in a reactive mode and potentially useful feedback data is not being gathered systematically. These problems necessitate the design of a new formalised improvement process that both, takes into consideration the characteristics of the clientele base and that operates in a pro-active mode. The design of the generic model improvement process is empirically driven. This implies that we must rst understand the current situation, analyse it, and based on the analysis, design an improvement process. The focus of this paper is on the part of the experiment leading to the characterisation of the clientele base, which is one result of the analysis. In this position statement we present the background for the experiment we are conducting, de ne the evolution paths of a generic model and describe the theoretical framework upon which the experiment is based. Following that we overview the experimentation process being performed and, conclude by presenting some of the early lessons learnt and a summary of the paper.