A practical taxonomy of data quality incidents

We sort two years of data incidents into six recurring shapes, then map each shape to a detection and response pattern. Bring one incident from your own backlog.