Quantifying value of adding inspection effort early in the development process: A case study

Abstract

Many researchers have reported the defect growth within the evolutionary-developed large-scale systems, and increased fault slips from the early verification stages into late. This suggests that improvement in the early defect detection process control is needed. This study focuses on evaluation of adding inspection effort early in the development process. Based on the examination of the existing metrics used in defect detection process, the authors establish metrics to quantify its value from the quality and cost-benefit perspective. The effect of adding inspection effort early in the development process is evaluated in a case study using industrial data from history and an ongoing project involving three geographically distributed sites of the same globally distributed software development organisation with around 300 developers. The findings show that the expert-based decision criteria for additional investment are mostly based on quality and reliability issues, and less on costs. Consequently, the additional inspection improves significantly the quality, while the cost-benefit was not statistically significant. This leads to the conclusion that better decision criteria that would incorporate the costs and not only quality perceptions are the key for improving the product reliability, as well as the overall software life-cycle cost-efficiency. This study is motivated by the real industrial environment, and thus, contributes to both research and practice by presenting the empirical evidence.