Aglet

Investigate Metric Grain Across a Report

A metric-grain investigation reconstructs one report number from a controlled set of rows. Include repeated events, shared entities, and an empty group where relevant. Compare the source, intermediate relation, aggregate, and displayed value so the brief can name the first grain change instead of blaming the final chart.

Build a useful investigation brief

  1. Build a grain fixture

    Create a compact dataset with one entity producing multiple events, two entities sharing a group, and a group with no event when the report permits it. Write the expected event, entity, and group measures separately. Record the report filters and period used for the observed number.

  2. Rebuild the number

    Capture rows after each join and filter, then record grouping keys and aggregate inputs. Compare the independently calculated measures with the report output and any export. Mark the first stage that duplicates, collapses, or drops the row unit the report is supposed to count.

  3. Test the leading explanation

    Change one grouping or join condition at a time and rerun the fixture. Check whether the disputed visual, card, and export move together. Treat a metric-definition conflict as a policy hypothesis until the owner confirms which grain the decision requires.

What to carry forward

The investigation is ready when the brief names the fixture, effective row grain, first divergent stage, affected report output, and strongest alternative explanation. Recommend a scoped metric or definition change. Leave mixed-grain cases open when no owner can select the decision unit.

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