Diagnosing Instagram DAU Drop
Question: Instagram sees a 5% decrease in Daily Active Users (DAU) over a week. How do you determine the root cause?
Recall: The GAME Framework
The GAME framework is a structured approach for analyzing product changes and issues:
- Goals – Clarify what the metric represents and why it matters.
- Actions – Identify user/system behaviors that may have changed.
- Metrics – Investigate supporting metrics that explain the shift.
- Evaluate – Isolate root causes and determine next steps.
Step 1: Goals
First, clarify what DAU means in this context—is it logins, feature usage, or full app opens? Then confirm whether the 5% drop is statistically significant by comparing against historical baselines and accounting for seasonality or holidays.
Also check: is the decline gradual or sudden, and is it localized to certain regions or global?
Step 2: Actions
Identify any recent changes in user behavior or system performance that could explain the drop:
- Was there a decline in app opens, feature usage (e.g. stories, reels), or content creation?
- Did push notifications fail to reach users?
- Was there a drop in content supply (e.g. creator output or virality)?
- Were there external events (e.g. outages, major news)?
Step 3: Metrics
To validate or eliminate hypotheses, drill into supporting metrics:
- Upstream and engagement metrics: Post views, likes, messages sent, time in app
- Platform health: Crash rate, latency, push delivery success
- User segmentation: Break down DAU by device, OS version, geography, user type (new vs. returning)
- Internal changes: Check launch logs for experiments, UI changes, or algorithm updates
- Support signals: Spike in customer complaints, app reviews, or Zendesk tickets
Step 4: Evaluate
Once you’ve gathered evidence, classify findings:
- Root cause: e.g. a push notification bug reduced re-engagement
- Contributing factor: e.g. fewer content creators posting that week
- Irrelevant variable: e.g. unrelated experiment with no reach
Summarize your findings into a timeline and prioritize fixes or mitigations. Communicate with product, engineering, and support teams to take fast action, then continue monitoring to confirm recovery.
clarify: so does the 5% drop a sudden drop or overtime in the one week does it broadly drop 5% or it dropped only in some regions or in some segments like new acqusition / frequent active customers? or does the 5% drop also happened last year same period? DAU = acqusition x activation x retention
segment: I will first quickly do some EDA to find out problem, like calculate the DAU drop in new customer, tenured customer, between regions to find out is there any difference. then I will also look at external factors like if competitors like tiktok has some new features just launched (check if they have surge in new customers or DAU) then I will check if there is current A/B testing on going and is the performance related to the DAU drop? or check if we recently have new ranking algorithm launched or new app version to see if there is causal relation in those.
make assumptions: if we assume the drop is due to app version we can conduct causal analysis by comparing customer updated app with customer not updated the app run a test like difference in difference or PSM, if we assume it is the ranking algorithm issue, then we will check our A/B test groups or run a switch back test, checking downstream engagement funnel metrics. If it’s external, I’d benchmark against competitors and historical seasonality.
Finally, I’d tie this back to business: a 5% DAU drop means fewer impressions and reduced ad inventory, which has revenue implications. I’d communicate findings quickly with product and eng leads, propose mitigation (rollback release, tune ranking, run re-engagement), and monitor recovery. The key is a structured approach that quickly distinguishes real causal factors from noise.