Diagnose a sudden KPI drop
Company: Meta
Role: Data Scientist
Category: Analytics & Experimentation
Difficulty: medium
Interview Round: Onsite
Overview: Diagnose why daily likes per active user came in 10% below its same-weekday baseline while active users stayed flat. Lay out data-quality checks, root-cause hypotheses, and a prioritized two-hour action plan. The hard part is telling a logging or pipeline fault apart from a real change in user behavior.