The model wins offline but loses online — which do you trust?

Decide from resultsMedium

Problem. A new ranking model beats the current one on offline metrics (NDCG, recall@k) but its live A/B shows flat-to-down satisfying engagement. Which do you trust, and what do you do?

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.

← Recs get more starts but lower completion — do you ship?Engagement is up week one — but will it retain? →