Both automated and manual testing are resource-intensive but essential components of quality assurance in software development. As system size and complexity increase, long test runs lead to high costs and delayed feedback to developers. While optimization techniques for automated testing are well established, there is a lack of empirical evidence for their transferability to manual testing processes in industry. This paper addresses this gap through a comprehensive empirical study involving five industry partners from different domains. We examine two language-independent optimization techniques—the established test impact analysis and the Pareto testing we have newly introduced—and evaluate their costs and benefits based on a historical analysis of 2,622 real bugs. Our results show that the techniques can be successfully transferred to manual testing, resulting in significant time savings and faster feedback. The practical relevance is underscored by the fact that all five participating companies have integrated one of the techniques studied into their testing processes.
Haas et al. (Thu,) studied this question.