Statistically Based Process Monitoring: Lessons from the Trench

Monitoring software processes is a non trivial task. Recently many authors have suggested the use of Statistical Process Control (SPC) for monitoring software processes, while others have pointed out its potential pitfalls. Indeed, the main problem is that SPC is often used "as is" without the appropriate customizations or extensions needed for making it applicable to software contexts. This work points out and discusses four main issues related to software process monitoring and highlights how SPC can be used as solution to address each problem. The solutions arise from experience collected by the authors during empirical investigations in industrial contexts. As so, this work is intended as a first step in clarifying how SPC can contribute to practically solve some monitoring issues and guide practitioners towards a more disciplined and correct use of the approach in controlling software processes.