Check whether the outcome improved
Exercise identity fixtures
Use repeated events, one entity in multiple groups, missing identifiers, and a reused key with known expected treatment. Verify global and per-group counts separately, including a period where an entity appears before and after the boundary.
Check transformed views
Apply filters, joins, drilldowns, and alternate groupings while keeping the identity rule fixed. Confirm each view counts the intended population and that overlapping groups do not silently present a summed count as a unique grand total.
Inspect labels and exports
Review metric names, definitions, tooltips, tables, charts, and exports for scope and null guidance. Ensure downloaded results preserve enough grouping and identity context for a reviewer to understand why totals need not add.
What to carry forward
Accept the measure when known identity cases match the declared rule across views, boundaries, and exports, with overlap and missing-key behavior understandable. Fail when a documented unique population is counted differently, or when labels imply additive groups that the evidence disproves.
Technical background: Metabase documentation.
Keep the decision with the work.
Use a Work Item in Aglet to record the problem, the evidence you have, and the next decision. Add an owner and priority, then keep updates in the discussion so the next person can follow the reasoning.
Create an account See the product workflow