Aglet

Learn from recurring distinct-count interpretation errors

Repeated distinct-count disputes reveal that teams often share a word such as users or orders while counting different populations. Learning should preserve the examples that expose overlap and missing-key behavior, then make the entity, scope, and additivity rule part of the report review vocabulary.

Keep the lesson for the next incident

  1. Cluster by identity question

    Group cases by entity key, null treatment, reused identifiers, overlapping groups, period resets, or join repetition. Record the calculation users expected and the population the report actually counted; keep valid non-additivity separate from defects.

  2. Write an identity standard

    For each measure family, state the counted entity, key, null policy, scope boundary, and whether group results are additive. Include a fixture with one entity crossing groups so the standard remains concrete rather than becoming a label alone.

  3. Make reuse reviewable

    Add a review prompt whenever a distinct measure is copied, joined, regrouped, or used in a ratio. Assign an owner for the identity rule and require cards, tables, and exports to retain the scope wording that makes future comparisons fair.

What to carry forward

The learning record is useful when it names the counted entity, preserves overlap and missing-key examples, and states whether grouped values add. If the organization needs two legitimate scopes, give them separate measures and definitions instead of preserving one ambiguous distinct-count label.

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