This path is for you if the p-value arrived before the decision. Parts cover a design that is more than a significance test, peeking and stopping rules, and a review meeting that actually decides. Start with the first part on this page. The stats series is the intuition. This one is the meeting.
- 1 How to design A/B tests that lead to good decisions, beyond the p-value p < 0.05 is not a product strategy. Experiment design beyond the p-value covers hypotheses, ethics, rollouts, and decisions that survive contact with real teams. Part 1 of Experimentation culture.
- 2 When to look at A/B test results and when to stop the test Stop peeking at interim p-values as final decisions. Use fixed horizons or sequential plans, guardrail metrics, and a written stopping rule before launch.
- 3 How to run an experiment review meeting that ends with a decision Run experiment reviews that decide: ship, iterate, stop, or gather one named fact. Use a short agenda, a one-page pre-read, clear roles, and a written decision log.