How would you design an A/B test for a new signup page?

Define the primary outcome, meaningful effect size, guardrail metrics, randomization unit, and planned duration before launch. Randomize eligible users consistently, avoid counting repeated visits as independent people, and verify instrumentation and allocation balance. Estimate sample size using a suitable baseline and statistical method. Analyze the assigned groups as intended and report uncertainty alongside effect size. Avoid repeatedly stopping when an ordinary p-value first becomes favorable. Check practical impact, accessibility, and segment risks before deciding whether to roll out the new page.