Your question is Use CUPED to Improve Experiment Power. Take a moment with it on the right.
Talk me through your thinking if you like. When you're confident, submit your answer and I'll grade it like a real screen (7/10 or better passes).
You work on a ride-sharing marketplace and are analyzing an A/B test where the treatment is expected to move a noisy outcome with strong pre-period correlation. The team is considering CUPED or another variance-reduction approach before deciding whether the experiment is conclusive.
How would you use CUPED or another variance-reduction method in the analysis, and how would you explain its impact on variance, power, and the final decision to ship?