Why departmental completion rates cannot simply be averaged
A group completion rate normally needs the group numerator and denominator, not an average of departmental percentages. Per-person metrics also need an explicit population and period.
On this page5 sections
Do not assume that averaging departmental completion percentages gives the group's completion rate. If the question is how much of the total target has been achieved, aggregate comparable actual and target amounts and divide them. An equal-weight average may answer a different question about the typical department, but it needs a different label. Quality rates, average order values and output per person each require their own numerator and denominator.
Prove the difference with two departments
In a hypothetical example, Department A has a target of 100 and actual output of 100, producing a 100% rate. Department B has a target of 900 and actual output of 450, producing 50%. Averaging their percentages gives 75%. The group has actual output of 550 against a target of 1,000, so its completion rate is 55%. Rounding did not cause the difference: the two calculations assign different weights. These are illustrative figures, not customer performance data.
When definitions align, weighting each departmental rate by its target is equivalent to dividing total actual output by total target. However, addition is not automatically valid when departments use different currencies, tax treatment, recognition dates or target versions. Resolve those boundaries first. Internal transactions also need the agreed treatment before their amounts enter a group result.
Choose the denominator for each measure
| Measure | Components to retain | Common mistake |
|---|---|---|
| Completion rate | Actual and target quantities | Giving every departmental percentage equal weight |
| Pass rate | Passing items and inspected items | Ignoring differences in batch size |
| Average order value | Amount and the agreed order count | Averaging daily averages without order counts |
| Output per person | Output and the relevant headcount or working time | Mixing period effort with month-end headcount |
Per-person measures need particular care. Part-time work, transfers, starters, leavers and people shared across departments may call for average headcount, full-time equivalents or recorded hours. The page developer should not choose among these without a business decision. Align deduplication and effective dates with the relevant personnel or operational records.
Handle departments without targets explicitly
A department with an explicit zero target and positive activity should show “Target is zero; attainment is not applicable,” rather than automatically displaying 0% or 100%. Use “No target assigned” only when the target is missing. If its actual amount contributes to the group numerator, explain that treatment. If it is excluded temporarily, disclose the excluded amount. A missing target, an intentional zero target and a failed interface are three different conditions.
Filtering out a department normally changes both numerator and denominator to the same set. A measure may deliberately retain a full-year group target, but its name must explain that choice. It then expresses contribution against that fixed target, rather than the completion rate of the selected departments. Test this distinction when a user changes organizational scope.
Keep calculations reproducible
Microsoft's SUMX documentation describes evaluating an expression for each row and summing the results in DAX. The function can implement certain weighted calculations, but it does not choose the business weight. Other implementations should likewise retain component quantities instead of storing only rounded percentages.
A display may use one decimal place while intermediate calculations retain the agreed precision. Recalculate a total from the underlying quantities; do not force it to equal the average of already rounded visible rows. An export should include the components, weight source and exclusion states needed for a business reviewer to reproduce the result independently. This is more useful than agreeing that two screens look approximately similar.
Test more than one company-wide number
Prepare cases covering large and small departments, zero and missing targets, shared staff, cancelled orders and date changes. Check detail rows, subtotals and totals. When they disagree, inspect the component quantities and record set before changing the formula. Record the rule in How to Build a Metric Definition Table for an Executive Dashboard; Which Metrics Belong on an Executive Dashboard? helps establish whether the chosen measure answers the management question.
The scope of data visualization development services should include these examples and calculation rules, not only a completion-rate card. When a department is added or a target changes later, determine whether the weight and definition version change before deciding whether to recalculate history. Preserve the original acceptance examples so that a maintenance change can be checked against the same expectations.