Using in-process metrics to determine the quality status of a software project under development is easier
said than done. How can you interpret a test-phase defect curve correctly to reflect the true quality status
of the project? If the defect curve is below a given baseline, is this a positive sign? What if the lower
curve is due to slow progress in testing? Likewise, how does one establish meaningful metrics for
design reviews and code inspections and interpret them correctly? How about metrics for stability and
reliability?
This paper describes the Effort/Outcome Model, which is a framework for establishing and interpreting inprocess
metrics in software development. The model has been validated and used on large scale
software projects in a mature software development organization. The central issue for in-process
metrics, the concept and definition of the model, and its use are discussed. Examples of metrics real-life
projects are provided.