Aglet

Investigate Cumulative Report Reset Logic

A cumulative investigation needs ordered values before and after a reset, two partition keys, a missing period, and a filter boundary. Recalculate the series independently and compare chart, table, card, and export results. The brief should distinguish ordering, partition, and reset policy rather than treating all divergences as arithmetic errors.

Build a useful investigation brief

  1. Construct ordered sequences

    Create rows for two groups across several periods, including a reset edge and a missing period where supported. Record expected running values, partition key, order field, reset condition, and filter scope. Keep a non-cumulative total as a control.

  2. Trace each accumulation

    Capture filtered rows, order, partition, cumulative inputs, reset detection, intermediate value, formatting, chart point, and export cell. Compare with the independent sequence. Mark the first point where a group inherits another group's value or a period starts from the wrong base.

  3. Test one boundary rule

    Change only partition, reset, missing-period, or ordering behavior and rerun the fixture. Compare later points and labels across surfaces. Treat carry-forward or restart preference as a policy hypothesis until the metric owner confirms the decision's intended horizon.

What to carry forward

The investigation is ready when the brief names the ordered fixture, first carryover or restart, affected series, strongest evidence, and tested policy alternative. Recommend a scoped accumulation or label change. Keep the case open when the reset event is not owned.

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