Talk Abstract: Product Experimentation

At Facebook controlled experiments (AB testing) are a core component in decision making. We use them not only to choose between alternatives, but also to prove or negate our core hypotheses. The importance of experiments and Facebook’s scale means we want all engineers to be able to run and analyze experiments, so the Data Scientist won’t be a bottleneck. This talk explains Facebook’s experimentation framework, and how we manage them in scale. It also discusses some more advanced considerations when designing such a system.

Bio: Data Scientist at Facebook