Design a Rigorous A/B Test for a Product Change

Shipping a change and eyeballing the numbers afterwards is not a test. This designs a proper A/B test: stated hypotheses, the sample size you need at 80 percent power, and the guardrails that stop you calling it early. Tell it the change, the segment and the primary metric, and you run something you can actually believe.

<context>
Your team wants to run an A/B test on [product change or feature, e.g. new onboarding flow / revised pricing page] for [user segment]. The primary metric is {PRIMARY_METRIC}.
</context>

<task>
**Design a statistically sound A/B test plan**:
1. State the null and alternative hypotheses
2. Calculate the required sample size (assume 80% power, alpha = 0.05; ask me for baseline conversion rate and MDE if not provided)
3. Define control and treatment variants precisely
4. List guardrail metrics that must not degrade
5. Specify the minimum runtime and any novelty-effect considerations
6. Describe how results will be analysed and what constitutes a ship/no-ship decision
</task>

<output_format>
- Numbered sections matching the 6 points above
- Include a simple decision matrix: outcome x action
</output_format>