Design A/B test for credit card offer

You are rolling out a new credit-card acquisition funnel that includes an updated APR disclosure and a signup bonus. Plan a complete A/B test: specify the randomization unit, eligibility and exclusion rules (such as current customers or fraud cases), primary and guardrail metrics (for example, approved accounts, activation rate, delinquency/default), the minimum detectable effect, statistical power, sample size, and how long the test should run. Cover selection bias (pre-approval, underwriting), interference across channels, peeking or early stopping, and regulatory limits (such as fair lending). How would you examine heterogeneous effects by segment while keeping false discovery under control? Overview: This question tests a data scientist's ability in experimental design, causal inference, metric selection and measurement windows, power analysis, bias reduction, identity resolution, and regulatory compliance for A/B tests on financial products.