Continuous Experimentation Empowers Software Development Decisions

Software experiments, adapted from the scientific method, help to establish causation and give predictive power. One of the most popular forms of experiment, called A/B testing, has been used repeatedly by technology forerunners like Google, Facebook and Microsoft over the last two decades. In A/B testing, two different versions of software can be tested with different sets of users without them even noticing. For instance, an old version of the software vs. a new version with a slightly changed color of a button in the user interface can be delivered to users. The collected user data, e.g. user clicks can be used to determine which version performs better, and product development decisions can be made based on the results.

Sezin Yaman’s PhD work focuses on how software experiments can be used to support development decision-making. Towards the end of her PhD studies, she did an HIIT sponsored research visit to Mozilla Corporation in Silicon Valley, California, where she worked on the applications of her current PhD research findings. During her stay, she conducted interviews and observational studies in order to explore how running experiments in an engineering organization can feed into creating better products.

The Mozilla visit was the last task in Yaman’s PhD project, and it greatly benefited her research work and future plans. At the moment, she is working on finalising her dissertation at the Computer Science Department, University of Helsinki, to be completed in the upcoming autumn. After that, she is planning to further explore software experiments and development decision-making processes in Finnish software companies.