Diagnose Facebook DAU Drop

Business Context

You own a dashboard for the Facebook app's core engagement metrics. On Monday morning, you see that global DAU dropped from 1.92B to 1.77B day-over-day (-7.8%), the largest decline in the past 12 months. Leadership wants to know by end of day whether this is a real user behavior change, a logging issue, or the result of a recent product or infrastructure change.

Over the same period, sessions per DAU fell from 4.6 to 4.1 (-10.9%), average session time was flat at 18.4 minutes, Feed impressions declined 6%, and ad revenue is down 5%. The drop appears concentrated in Android (-12%), while iOS is down 2% and Web is flat. A new Facebook Feed ranking update launched to 35% of users two days ago, and the Android app version 462.0 reached 48% adoption yesterday. Data engineering also notes a delayed backfill in one event pipeline from APAC between 2am-6am UTC.

Requirements

Define the DAU metric precisely, including what counts as an active user and how you would validate the definition.
Lay out a structured investigation plan to determine whether the decline is caused by instrumentation, data latency, product changes, app version issues, or real behavior shifts.
Decompose the metric into the most useful cuts and identify which cuts you would check first.
State 3-5 concrete hypotheses for the drop and explain how you would test each one.
Recommend immediate next steps for stakeholders, including what you would communicate before root cause is fully confirmed.

Data Available

facebook_app_events: user_id, event_time, event_name, app_version, platform, country, surface
facebook_sessions: user_id, session_id, session_start, session_end, platform, app_version
feed_impressions: user_id, impression_time, story_id, ranking_model_version, surface
experiment_exposure_log: user_id, experiment_name, treatment, exposure_time
etl_pipeline_monitoring: pipeline_name, region, expected_rows, actual_rows, delay_minutes, run_status
release_rollout_log: app_version, platform, rollout_percent, release_time

Business Context

Requirements

Define the DAU metric precisely, including what counts as an active user and how you would validate the definition.

Lay out a structured investigation plan to determine whether the decline is caused by instrumentation, data latency, product changes, app version issues, or real behavior shifts.

Decompose the metric into the most useful cuts and identify which cuts you would check first.

State 3-5 concrete hypotheses for the drop and explain how you would test each one.

Recommend immediate next steps for stakeholders, including what you would communicate before root cause is fully confirmed.

Data Available

facebook_app_events: user_id, event_time, event_name, app_version, platform, country, surface

facebook_sessions: user_id, session_id, session_start, session_end, platform, app_version

feed_impressions: user_id, impression_time, story_id, ranking_model_version, surface

experiment_exposure_log: user_id, experiment_name, treatment, exposure_time

etl_pipeline_monitoring: pipeline_name, region, expected_rows, actual_rows, delay_minutes, run_status

release_rollout_log: app_version, platform, rollout_percent, release_time

Problem

Business Context

Requirements

Data Available

Diagnose Facebook DAU Drop

Problem

Business Context

Requirements

Data Available