Towards meaningful benchmarking of software development team productivity

Abstract

Software development projects are known for inaccuracies associated with elapsed time and total cost estimates. Attempts have previously been made to provide tools to facilitate estimation of just how much effort will be required. One such tool is the estimation of project size (and therefore effort and time required) using function point counts. This benchmarking tool facilitates measures of productivity relating this size to the person‐hours required. The problem with this is that such relative productivity measures assume labour hours to be homogenous and that the only measure of output is the size of the project. This paper investigates the use of data envelopment analysis as a method of benchmarking which overcomes these issues. The end result is a set of simple tools that can be used to determine whether a given project or project plan is efficient.