You are planning an A/B test for a new in-app feature in a ride-hailing product. The current user-level conversion rate on the primary metric is 18.0%, and you want to detect an absolute lift of 1.2 percentage points with 80% power at a 5% two-sided significance level using a 50/50 traffic split.
How many users do you need in each variant to reliably detect the target effect, and what total sample size should you plan for?