The broken source.
The cascading damage.

Row count anomaly detected in source table. Expected ~2,800 rows, found 1,400.

4 downstream models affected: stg_orders, fct_revenue_realtime, dashboard_realtime_metrics, customer_metrics_view

• stg_orders — Rebuilds on partial source data
• fct_revenue_realtime — Missing order records will show undercount
• dashboard_realtime_metrics — Real-time charts display incomplete totals
• customer_metrics_view — Customer LTV and retention metrics affected

The ingestion job completed successfully, but the upstream data source is missing orders. Check: (1) Did the orders API experience an outage? (2) Are you filtering orders by date, and did the time window shift? (3) Was a recent ETL change applied?

• Verify raw data in source system (check API/database directly)
• Compare last successful ingestion timestamp and row count
• Review recent changes to raw_orders dbt model or upstream ETL
• If source is actually correct, pause downstream models until decision is made
• Once root cause is fixed, manually re-run stg_orders and downstream models

• 1. Source anomaly detected — raw_orders row count (1,400) vs. baseline (2,800)
• 2. Lineage traced — Found 4 downstream dependencies
• 3. Impact assessed — All 4 models will produce incorrect results
• 4. Grouped into single incident — Prevents alert fatigue from 4 separate Slack messages
• 5. RCA hypothesis generated — Ingestion success but upstream data loss suspected