Retention is already 92% — can your test even detect a change?

Power & sample sizeHard

Problem. You want to test whether a feature improves WhatsApp retention, but D30 retention is already ~92% and barely moves. How do you design a test that can actually detect an effect, or decide you can't?

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.

← A messaging feature spreads person-to-person — how do you randomize?A growth feature might raise the block rate — which test, and can you see it? →