Tag: treatment-effects

Engaged users are often different before an intervention begins. Causal inference helps separate “the feature worked” from “the feature found users who were already likely to succeed.”