You can't read messages — how do you pick a metric to experiment on?

Metric designHard

Problem. You want to A/B test a new WhatsApp feature that should make conversations "better," but you can't read message content. How do you choose a valid experiment metric, and how do you know it actually reflects the outcome you care about?

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? →