A survey on software estimation in the Norwegian industry

Abstract

We provide an overview of the estimation methods that software companies apply to estimate their projects, why those methods are chosen, and how accurate they are. In order to improve estimation accuracy, such knowledge is essential. We conducted an in-depth survey, where information was collected through structured interviews with senior managers from 18 different companies and project managers of 52 different projects. We analyzed information about estimation approach, effort estimation accuracy and bias, schedule estimation accuracy and bias, delivered functionality and other estimation related information. Our results suggest, for example, that average effort overruns are 41%, that the estimation performance has not changed much the last 10-20 years, that expert estimation is the dominating estimation method, that estimation accuracy is not much impacted by use of formal estimation models, and that software managers tend to believe that the estimation accuracy of their company is better than it actually is.