You own the safety guardrail on another team's engagement test

Metric designHard

Problem. A ranking team is running an A/B test of a change that boosts engagement. Your job is to provide the safety guardrail: a metric that flags if their change also quietly increases how much harmful content users are exposed to. They don't want to be blocked on noise, and you don't want a real harm regression to slip through. How do you design this?

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 safety filter should cut a rare harm — but prevalence is too rare to move in the testA new enforcement model cuts harm but over-blocks good users — ship it? →