Meadow Operations prepares a report for another analyst. The receiving analyst needs predictable field names and types, along with separate evidence of duplicate identifiers and missing matches. Preserve those concerns as distinct outputs instead of folding an exception into a measure where it will be hard to explain.
The tasks revisit Select, Unique, Join, Filter and Summarize. Some exception rules return an empty table, which can be a valid checked answer. Check the business condition rather than adding records to make the result look more useful. For schema tasks, a correctly spelled name and a text identifier matter as much as the visible amount. The hidden extract changes the inputs so your report must state a repeatable transformation.