Back to Meta
Meta · Product AnalyticsFree

Facebook Stories DAU dropped 15% — walk me through your investigation.

metricsroot causeDAUdiagnosis

Question

You are the PM for Facebook Stories. You wake up on a Monday morning and see that Stories DAU (Daily Active Users) has dropped 15% week-over-week.

Question: Walk me through your investigation. What do you check first? How do you find root cause? What do you do?

Step 1: Don't Panic — Validate the Data

Before investigating causes, confirm the signal is real. Check: Is the logging pipeline healthy? Are there any instrumentation changes or A/B test rollouts that could affect measurement? A 15% drop in one dashboard metric is serious, but I've seen it be a data pipeline issue 30% of the time.

Step 2: Characterize the Drop

Once confirmed real, I'd answer: When did it start? (Gradual or sudden — sudden suggests a bug/rollout, gradual suggests product or competition). Who is affected? Segment by: platform (iOS vs Android vs Web), geography, user type (creators vs viewers), age group. What metric family is it? Is it creators dropping off, viewers dropping off, or both?

Step 3: Form Hypotheses by Segment

  • If iOS only → recent iOS app update or Apple policy change
  • If a specific country → local news event, outage, or competitor launch
  • If creators only → something changed in the creation flow or camera experience
  • If all segments → likely a feed algorithm change or feature rollout

Step 4: Investigate

I'd cross-reference with: recent code deploys, A/B test assignments, any major competitor announcements (Instagram Reels update, TikTok changes), external events. Talk to eng + data science immediately.

Step 5: Action

If a recent rollout correlates — consider rollback while investigating. If external — monitor for recovery. In all cases, set up a 24h war room cadence until root cause is identified.

Meta Expectation

Meta interviewers want to see: structured thinking, segmentation instinct, hypothesis-first approach, and urgency calibration. Don't jump to solutions before you understand the problem.