Identifying the causes of poor progress in software projects

Abstract

We present data on the progress of two projects at IBM Hursley Park. Each project lasted approximately 12 months in duration. We use the data to identify the areas in the projects where poor progress was occurring and to investigate the causes of this poor progress. We find some similarities between the two projects in terms of where some poor progress is occurring i.e. with the design, code and test processes. But we also identify differences between the two projects that can be partially explained by the dependency of these two projects on other parts of IBM. We also find that limited quantitative data is reported in the projects, and that there is little explicit comparison of actual progress with planned progress. Indeed, the reporting of progress seems to be affected by factors like the 'deadline effect' and preferences for reporting certain types of progress. We conclude that these factors may threaten the valid reporting and management of the projects.