Data-driven fashion systems of the future will revolutionize the way consumers shop for clothing and choose outfits: imagine an automated personal stylist that ships clothes straight to your door based on their compatibility with your existing wardrobe, the upcoming events on your calendar, and style trends learned from the web. To build such systems, we must identify the fashion activities that are the largest consumer pain points, the interventions necessary to alleviate those pains, and the computational models that enable those interventions. To guide the design of these next-generation tools, we propose an experimentation engine for fashion interfaces: leveraging social media platforms to run multivariate design tests with thousands to millions of users. Social platforms are already home to dedicated communities of fashion enthusiasts, and expose programmable agents - chatbots - that can be used to rapidly prototype data-driven design interfaces. Measuring the number of followers and user engagement amongst these prototypes can inform the design of future standalone fashion systems. At this workshop, we will sketch the design space of fashion experiments, and present preliminary results from deploying our "fashion bots."