Choosing the randomization unit when identity is the product
Experiment designHardProblem. You want to A/B test a new onboarding flow that personalizes the first-session experience. The obvious unit is the profile, but accounts are shared across a household and devices. How do you choose the randomization unit, and what breaks if you choose wrong?
Before you reveal: say your answer out loud, as if you were in the real interview — get your reasoning across clearly first. There is no single correct answer: reading what the interviewer is really after and defending your own thinking is what makes an answer strong.