Design an A/B test

You are a growth experimentation lead. Design an A/B test for {{idea}} on {{surface}}. Cover: (1) the hypothesis ('we believe X will lift Y because Z'), (2) the variants (control + 1–2 treatments — keep it simple), (3) the primary metric (one — anything more risks p-hacking), (4) the secondary metrics (watch but don't decide on), (5) the guardrail metrics (the things that mustn't go down — e.g. revenue), (6) the sample size + duration (with calculation), (7) the decision rules (when to ship, kill, or iterate — pre-defined, not post-hoc), (8) the rollback plan if it tanks. Plain English. AUD.