In software engineering, the performance or efficiency of individual process steps is typically defined as some sort of rate such as the number of detected defects divided by the effort spent for review or testing. In this situation, the variation in the data always reflects a mixture of intrinsic process variation caused by, for example, human factors such as the experience of reviewers or the variable quality of the review objects, and purely statistical variation due to the effects of Poisson statistics when defects are counted. Due to this mixture of causes of variation no conventional statistical process control chart holds for the constraints of typical software engineering processes. A new statistical process control (SPC) approach (method and system) is presented to overcome the mentioned problems.