Hypothesis

A hypothesis is a specific, falsifiable claim about user behavior that an experiment or analysis is meant to confirm or reject.

A good hypothesis is not a wish but a structured sentence: if we make this change, this metric will move by this much, for this reason. Its value is that it is written before the data is seen, otherwise the mind recasts any outcome as having been predictable after the fact. It has two common traps: a hypothesis with no clear metric that can never be rejected, and one that bundles several changes so it is unclear afterward which part had the effect. A hypothesis needs a single target metric and a defined direction so the experiment result stays interpretable.