Delivering user-centric software features becomes a core competence in all business areas. Minor software changes can have a significant user impact, even if the effects are not foreseeable. Thus, continuous experiments, which are based on experiment specifications, launch two variations of a software to collect information about user behaviour in production and evaluate the recorded measurable changes.
The goal is to develop a software artefact that simplifies and ultimately improves the quality of this testing process. Therefore, the tool needs to evaluate the given experiment specification (JSON) based on syntactical and logical errors. Afterwards, the tool automatically extracts the needed information from the given experiment specification to synchronize the experiment with the continuous experimentation platform.
To evaluate the implemented tool, a proof of concept with a website providing two different recommenders is executed.