AB Testing Emphasis

ABtest 了解多少，哪个环节比较重要，如何评估分流的好坏，是否够随机

Asked in the round 1 stage. Explain how you would design and evaluate an A/B test end to end, with particular attention to whether users are assigned randomly and whether the split is trustworthy. No product-specific context, traffic volume, baseline metric, or test duration was provided, so state the assumptions and additional inputs you would request.

1. State the null and alternative hypotheses, primary metric, guardrails, and a meaningful MDE.
2. Choose the randomization unit and explain why it is appropriate.
3. Describe how you would assess allocation quality, including sample ratio mismatch and balance checks.
4. Show a sample-size calculation using clearly labeled illustrative assumptions, then explain how traffic and duration determine feasibility.
5. Define the analysis, monitoring, and ship or do-not-ship decision rules.