Decide launch with p=0.10 at alpha 0.05

An A/B test was pre-registered with α=0.05 (two-sided), power=80%, and MDE=+2% relative lift in 30‑day subscriber retention. After 4 weeks, the observed lift is +1.4% relative with p=0.10; guardrails show no harm. False-positive cost is estimated to be 3× the false-negative cost. (1) Give a binary launch/no‑launch decision under the pre-registered frequentist plan and justify. (2) Compute or outline the conditional power and additional sample size/duration needed to achieve 80% power at a 1.4% MDE; recommend extend vs stop. (3) Reframe the decision with a simple Bayesian analysis (e.g., Beta‑Binomial or normal approximation) using a weakly informative prior; compare expected loss for launch vs no‑launch under the 3:1 cost ratio. (4) If choosing to ship despite p=0.10, propose a risk‑mitigated rollout (e.g., phased ramp, sequential testing/CUPED, kill switches) and explicit rollback criteria.