A Follow-Up Reflection on Software Process Improvement ROI

Our discipline must shift toward value-based software engineering, because we're obliged to prove our contributions to the financial bottom line. In the May/June 2004 IEEE Software special issue on return on investment (ROI), the author presented measurement results for the ROI of software process improvement (SPI). This article made three main contributions. First, provided a detailed overview of publications containing real-life measurement results from practical applications of SPI, in which the author measured the ROI. My study included 20 cases, with an average ROI of 7 and a median of 6.6. This indicates that SPI's net profit seems to be approximately US$7 for every dollar invested. However, I found no published cases in which SPI investments resulted in a measurable loss; furthermore, the ROI bandwidth was large (between 1.5 and 19). This indicates that the actual ROI of an SPI investment seems hard to really guarantee up front. Second, I showed that benefits are just as easy to quantify as costs. Cost measurements are always based on an agreement about how to measure and quantify costs. Such an agreement can also serve as the basis for measuring benefits. My article contained data from two real-life projects that had made such cost and benefit measurements and calculated ROI. Finally, I concluded that expressing "value" is crucial. Software engineering and its improvement are often major investments for organizations. Investments must be profitable. Because different people in different roles share one generic term for value-money, I recommended expressing any software engineering effort and its benefits in financial terms.