Marco Polo - a tool for automated exploratory testing of previously unseen online stores

Abstract

Online stores continue to increase in popularity with 2.3 billion people estimated to have shopped online during 2022. Whilst every online store is unique in its own right, the online shopping domain itself is quite constrained in terms of the type of functionality being offered. This begs the question as to why companies invest so much time, effort and money into developing complete test suites for their specific systems. In this paper, we argue that it would be more efficient for companies to leverage the abundant common ground between most systems in the domain such that they need only focus on features that make their product unique when it comes to allocating testing effort. We go on to present Marco Polo, an automated exploratory testing tool inspired by the renowned explorer and trader from the 13th century. Using behavioural cloning to train on expert traces from a sample of online stores, this tool is able to exploratory-test previously unseen websites whilst handling reactive websites and events such as cookie/GDPR consents. We discuss design decisions as well as challenges and opportunities for further development. A demo URL is also provided.