The aim of verification and validation is to ensure the quality of a software product. The main fault detection techniques used are software inspections and testing. Software inspections aim at finding faults in the beginning of the software life-cycle and testing in the end. It is, however, difficult to know what kind of faults that are found, the severity of these, how much time it will take to correct them etc. Hence, it is important for a software organization to know what faults that can be found in inspections and testing, respectively. This paper reports on a case study over 2 years in a large project.

The purpose of the case study is to investigate the trade-off between inspections and testing. A measure of goodness is introduced, which measures if faults could have been found earlier in the process. The measure was successfully used to illustrate the effect on when faults are found concerning a process change, a project decision, and extensions of developed test cases. An increased focus on the development of low level design specifications before the coding activity, and testing improvements in early phases were concluded to be important process improvements for the organization.

The results also concern how much resources that are used in the various development phases, from requirements development to system verification, and the severity of faults that are found in the various phases. The results will be used as input to a quality improvement program in the company.