Check whether the outcome improved
Define grain assertions
Write the expected source row unit, grouping keys, distinctness rule, and measure for each fixture. Include repeated events, shared groups, and a no-member group where supported. A pass requires the report label and the calculation to describe the same unit.
Exercise each surface
Apply the same scope to the card, table, chart, and export. Compare row membership, aggregate inputs, displayed total, and exported value. Include a drill or filter that could change grouping, and record whether the grain remains stable or is intentionally different.
Review reader clarity
Ask a reviewer to identify the unit from the report title, measure label, nearby explanation, and export header. Preserve the fixture, query or transformation evidence, and report revision. Record a bounded result when one surface lacks enough context to establish grain.
What to carry forward
Verification passes when the documented row unit produces the expected measure across reviewed surfaces and the reader can identify that unit. Report a bounded pass for uncovered drill states. Keep the result inconclusive when a card, chart, or export silently changes grain.
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