Diagnose sudden KPI drop with segmentation

On 2025-09-01, daily Likes actions on a global social network fell 10% versus the prior 14-day average, while DAU stayed flat. Treat this as a production incident. Outline an end-to-end diagnostic plan: 1) enumerate concrete data-validation checks you will run (e.g., event volume parity vs app_sessions, null-rate and schema-change diffs, client→server receipt ratios) and what thresholds scream instrumentation failure; 2) rule out expected variability by checking seasonality, regional holidays, release calendars, and major outages/events—what external data would you join and how; 3) localize the drop by funnel stage (feed impressions → post views → like impressions → likes) and by segments (surface: friends vs pages vs events; platform; app version; geography; new vs retained users); 4) state three falsifiable hypotheses and the exact cuts/queries/plots to test each; 5) define the decision tree that gets you to a root cause within 24 hours, including what you’d do if multiple segments move in opposite directions; 6) propose one immediate mitigation and one follow-up experiment to prevent recurrence. List the first three queries or charts you would run and the precise outcomes that would make you conclude behavior change (not logging) is at fault.