Usability Improvement Through A/B Testing and Refactoring

• Autores: Firmenich, Sergio; Garrido, Alejandra; Grigera, Julián; Rivero, José Matías; Rossi, Gustavo Héctor
• Año de publicación: 2019
• Idioma: inglés
• Tipo de recurso: artículo
• Materia: Ciencias de la Computación e Información; Usability evaluation; A/B testing; Web refactoring; Agile methods; External quality
• Institución: Comisión de Investigaciones Científicas de la Provincia de Buenos Aires

Usability evaluation is an essential task in web application development. There have been several attempts to integrate user-centered design with agile methods, but it is hard to synchronize their practices. User testing is very valuable to learn from feedback of actual use, but it remains expensive to find and solve usability problems. Furthermore, the high cost of usability evaluation forces small/medium-sized companies to trust the first solution applied, without actually testing the success of the solution or considering a possible regression in usability, as could be highlighted by an iterative testing method. In this article we advocate for a usability improvement cycle oriented by user feedback, and compatible with an agile development process. We propose an iterative method supported by a toolkit that allows usability experts to design user tests, run them remotely, analyze results, and assess alternative solutions to usability problems similarly to A/B testing. Each solution is created by applying client-side web refactorings, i.e., changes to the web pages in the client which are meant to improve usability. The main benefit of our approach is that it reduces the overall cost of user testing and particularly, A/B testing, by applying refactorings to create alternative solutions without modifying the application’s server code. By making it affordable for usability experts to apply the method in parallel with the development cycle, we aim to encourage them to incorporate user feedback and try different ideas to discover the best performing solution in terms of the metrics of interest