To enable data analysts and research scientists in the Foundation to perform cross-wiki, cross-platform randomized controlled experiments, longitudinal studies, multivariate A/B tests, and adaptive experiments (e.g. Bayesian optimization and multi-armed bandits) on a centralized experiment platform. Part of this will be evaluating whether we can utilize existing services or whether – due to privacy concerns and complexity of our infrastructure – we will need to roll our own.

Inspired by: Building our Centralized Experimental Platform blog post from StitchFix