One branch shows average preparation time of five minutes and another fifteen. Should the dashboard show ten minutes when both are selected? That depends on whether the question gives branches equal weight or asks for average time per order across all orders. The number alone does not state the definition.
Before accepting a dashboard, select an important metric and prepare a sample with known results. The business intelligence deliverables guide covers overall scope. Here the focus is a calculation acceptance sheet exposing differences in averaging, filtering and missing cases rather than visual design.
A sample revealing aggregation differences
Consider a teaching example of internal order preparation. The first branch completes ten orders taking five minutes each; the second completes ninety taking fifteen minutes each. Assume complete time information and no excluded orders for this sample only.
| Group | Orders | Total preparation minutes | Average per order |
|---|---|---|---|
| First branch | 10 | 50 | 5 minutes |
| Second branch | 90 | 1350 | 15 minutes |
| Both branches | 100 | 1400 | 14 minutes |
For average time per order across both branches, divide 1400 minutes by 100 orders to obtain 14 minutes. Adding the two branch averages and dividing by two gives 10 minutes, assigning branches equal weight despite different order counts. That may answer another question, but should not silently carry the same average-order label.
This is an original acceptance example, not a suggested preparation benchmark or real business result. Ask the implementer to reproduce it from sample data and reveal what contributes to the total. Do not accept manually changing a number to match the table: the rule must also work when data changes.
Define the measured operation
What starts and ends preparation? Are known waiting periods included, and which unit does the source use? Document the approved definition, selection date field and record level representing one order. Multiple order lines should not unintentionally increase an order's weight.
Define treatment of cancelled, incomplete and missing-timestamp records. These are meaning decisions before formulas. Hiding missing cases does not eliminate their effect. Where necessary for interpretation, dashboard users should understand excluded case counts and reasons.
Six acceptance-sheet cases
- Each branch separately: reconcile the result, count and total minutes with manual sample calculations.
- Both branches: apply the approved weighting definition rather than unexplained averaging of displayed values.
- Period filter: include or exclude orders under the specified date field, including a boundary case.
- No eligible records: show the agreed outcome and absent denominator without confusing it with zero duration.
- Missing time or excluded state: apply the approved decision with an explanation of its effect.
- Multi-line order: avoid repeating duration or count when the metric is defined per order.
For each case, record data, filters, expected and actual results, discrepancy reason and tested version. Change a known sample value and repeat. This tests calculation and filtering together rather than comparing screenshots potentially based on different record groups.
Filters are part of the number's meaning
Power BI documentation explains that measures change with report interaction and model relationships propagate filters between tables. Selecting a branch may therefore affect other data groups under the design. Request a practical explanation for your metric rather than a general statement that filters work.
Select a branch and period in the example, then inspect the records explaining the average through an authorized method. If the denominator and minutes come from different sources, ask about their relationship and filter scope. Do not attribute differences to rounding before confirming identical underlying populations.
Receive the definition and change authority
Attach the metric name, definition version, data source, last read time and meaning approver to the acceptance sheet. Record display rounding without hiding underlying calculations needed for review. When exclusions or date fields change, identify tests to repeat and effects on historical comparisons.
Power BI data readiness addresses inputs before construction. When requesting a dashboard from Nextwave, use the sample and case sheet to define acceptance. Colors and alerts become useful when the team can explain the number prompting a decision.
References and preparation
The Power BI measures guide and model relationships guide were opened on 4 November 2026. They support calculation and filter effects, not one universal business definition. Samples and cases are editorial; AI assisted preparation with arithmetic, meaning and source review.
