Companies like Spotify often use A/B testing to make product decisions that will drive key company KPIs (e.g. user engagement, revenue). Spotify recently decided to retire its A/B testing platform and build a new experimentation platform that reduces the time to modify or create new experiments, decreases the volume of events logged, standardizes analysis, and assigns users to experiments more efficiently. The new system enables Spotify to assign users to experiments through remote configuration (wherein they change the experience a user receives based on “properties” of the client). It also leverages a Metrics Catalog to manage, store, and query metrics and an Experiment Planner (within Backstage) to create, launch, and stop experiments.