1000s of experiments and daily peeking — control the error

Multiple testingHard

Problem. Bing runs 1000s of experiments at once, each on dozens of metrics, on dashboards people refresh daily. Two failure modes: false discoveries across all those tests, and peeking. How do you control error at org scale?

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.