The A/B Test Designer

Design an A/B test to answer this question reliably.

What I want to test: [DESCRIBE THE CHANGE]
The metric that matters: [PRIMARY METRIC]
My traffic/volume: [DESCRIBE]
Current baseline: [CURRENT RATE/VALUE]

Provide:
1. The hypothesis and the primary metric (one, not ten)
2. The control and variant, and what to hold constant
3. The sample size and run time needed to detect a meaningful effect
   (specify the minimum effect worth detecting)
4. The guardrail metrics to make sure I'm not winning locally while
   losing globally
5. The traps (peeking, stopping early, seasonality, novelty effects)

Give me a test I can actually run and trust, not one that produces a
falsely confident result.