Design robust A/B test with interference and seasonality

You are launching a redesigned onboarding flow expected to increase Day-7 activation but may cause network effects (users invite others) and weekly seasonality. Design an experiment plan that covers: (1) hypothesis, primary metric(s), guardrail metrics, and exact metric definitions with attribution windows; (2) unit of randomization and exposure (user, household, geo, or cluster) and why, given potential interference; (3) sample size and power analysis, target MDE, duration assumptions, and how you’ll account for seasonality (e.g., full-week multiples); (4) variance reduction (e.g., CUPED with pre-period covariates), stratification, or geo-matched pairs; (5) SRM detection and remediation plan; (6) sequential monitoring approach and stopping rules (alpha spending) to avoid p-hacking; (7) a ramp plan with holdouts and a plan for novelty/learning effects; (8) diagnostics for noncompliance, bot traffic, and triggered vs. assigned populations; (9) how you’d detect and mitigate interference/spillovers (cluster randomization, geo experiments, or switchback) and quantify any bias if user-level randomization is used; (10) interpretation plan if primary and guardrail metrics disagree, and how you’d decide to ship.