Diagnosing and recovering a 10-day fall in Facebook engagement

You are a product manager for Facebook. Over the last 10 days, user engagement has declined significantly. Explain how you would find the cause and what you would do about it. ```hint Let the 10-day window guide you A decline that began recently usually traces back to something that changed around a specific date. Confirm the decline is real, find where it is concentrated, and then line it up against what changed inside and outside Facebook at that time. ``` ### Clarifying Questions - Which engagement metric fell, and by how much: daily active users, time spent, sessions per user, or actions such as posts, comments and reactions? - Did the decline start abruptly on one day or build gradually over the 10 days, and has it recovered at all since? - Is it visible on iOS, Android and the web alike, and in every region, or concentrated somewhere? - What is the comparison baseline: the 10 days before, or the same period last year? ### What a Strong Answer Covers - Pinning down the metric, the size and the start date before diagnosing - Ruling out logging and data-pipeline problems with independent signals - Segmenting by platform, app version, region, user cohort and surface to localize the decline - Internal and external hypotheses ordered by fit with the evidence, each with a check that would confirm or reject it - A response covering the fix, communication to stakeholders, and monitoring that catches a repeat sooner ### Follow-up Questions - If time spent per user fell but daily active users held steady, how does your list of hypotheses change? - Suppose the decline lines up with a Feed ranking change that improved its own target metric in testing. What would you do? - How would you separate the effect of a holiday or a major news event from a product problem?