When a pipeline loads bad data, the damage spreads quietly into dashboards, invoices and models before anyone notices. This eight-page document gives a data platform team one place to record what broke, which datasets and consumers were hit, why the system allowed it, how the backfill ran and what will change.

The sample follows Everdale Labs through a renamed field that loaded empty revenue values for 3 hours 40 minutes. Replace the incident ID, UTC times and summary first, rebuild the timeline from your paging tool, then swap in your own causes and give every action item one owner and one due date.