Benchmarking usability of early designs using predictive metrics

This paper introduces a set of novel metrics-based prediction models that allow stakeholders and user interface designers to assess, compare and choose among alternative designs in early development stages. Most of the existing usability evaluation methods require a fully functional prototype. Tests are also mostly conducted after the development and deployment of the software. Tests also require a specific costly lab which may result in significant maintenance costs. The proposed models in this paper reply on the correlations that we discovered between predictive usability metrics and the results of usability tests performed by users. Our empirical investigations show that predictive metrics and user-oriented tests conducted by users provide similar scores regarding the overall usability as well as other parameters such as learnability and efficiency of alternative designs before their development and deployment.