Root-Cause Analysis for a Sudden 10% Overnight Fall in Facebook Event RSVPs
Company: Meta
Role: Product Manager
Category: Analytics & Experimentation
Difficulty: medium
Interview Round: Other
You are the product manager for Facebook Events. RSVPs to events fell by 10% overnight. How would you work out what caused the drop?
```hint Read the shape of the drop
A one-day step change points to different causes than a slow slide. Think about which cuts of the data (platform, app version, event type, response type) would separate a tracking or release problem from a real change in what people do.
```
### Clarifying Questions
- What does the RSVP metric count: every response, or only "Going" and "Interested", and is it a raw daily total or a rate per invitation or per event view?
- What is the drop measured against: the previous day, the same weekday last week or a forecast, and how large is the metric's normal day-to-day swing?
- Is the drop spread evenly, or concentrated in one platform, app version or country, or in public rather than private events?
- Did anything change around the time of the drop: an app or web release, a change to how events are ranked or recommended, a logging change or an experiment ramp-up?
### What a Strong Answer Covers
- A precise metric definition and baseline, and a check that the drop is real rather than a data or logging artifact
- Segmentation by platform, app version, region, event type and response type to localize the drop
- Breaking RSVPs into steps (events available, events seen, event pages opened, RSVP taps, responses saved) to find which step moved
- Internal and external hypotheses ranked by how well they fit an abrupt one-day change, each paired with the data that confirms or rules it out
- What to do once the cause is found, and how to confirm the metric recovers
### Follow-up Questions
- The drop is limited to the newest iOS app version, yet RSVP taps per event page view look normal. What do you check next?
- A new event-recommendation model launched the same day: event page views went up while RSVPs fell. How do you decide whether to roll it back?
- What monitoring would catch and localize a drop like this within hours next time?
Overview: A product metrics diagnosis question for a Facebook Events product manager who sees RSVPs fall 10% overnight. It tests defining the metric and baseline, ruling out data problems, segmenting by platform and event type, breaking down the RSVP funnel, and ranking hypotheses with a test for each.