Gains from improved recruitment of software developers

This paper is about how much software developers vary in their productivity and how much companies can benefit from improving their recruitment processes. I find that the shape of the productivity distribution defends a stronger focus on avoiding selecting low-productivity developers when recruiting for the completion of a particular task or project. When, on the other hand, hiring developers for a permanent job, stronger emphasis should be put on identifying those with high productivity. I develop a model that predicts the expected value creation from improved recruitment processes. The model is used to calculate the gains of going from a CV and unstructured interview-based to an improved recruitment process. Finally, I provide recommendations on how to improve the recruitment process based on results from recent meta-studies on how skill tests and other recruitment process elements correlate with actual work performance.