You are asked which statistical methods you find most useful for product growth analysis. To make the discussion concrete, suppose you are evaluating a new in-app rider coupon banner: among 20,000 exposed users, 2,360 took a ride within 7 days, versus 2,180 of 20,000 unexposed users. You also have user-level covariates such as prior 30-day ride count, city, and device type, and you want to decide which methods are most appropriate for measuring impact.
How would you analyze this problem statistically, and which methods would you prioritize to estimate whether the banner truly improved ride conversion?