Why dashboard growth rates look wrong: incomplete periods and zero baselines
Compare like-for-like periods while a month is still open. A zero baseline should not produce an invented growth percentage; show the period, baseline status and data cutoff.
On this page5 sections
When a dashboard shows an unexpected growth rate, check the comparison window before checking the chart. A partial month compared with a complete previous month answers a different question from a like-for-like comparison. A zero or negative baseline also needs an explicit interpretation. Displaying a percentage and a red or green arrow does not make the result meaningful.
Write down the dates before choosing a formula
Define the business date, time zone, start and end boundaries, cutoff time and completeness status. An order date is not interchangeable with a delivery or payment date. If the current source is complete only through yesterday, decide whether the reference period should end at the corresponding point. Including today's unfinished activity can introduce another mismatch.
Consider a hypothetical example in one consistent unit: revenue is 120 for the first ten days of this month, 100 for the first ten days of last month, and 360 for the whole previous month. The equivalent-progress increase is 20%. Comparing 120 with 360 produces a decline of about 66.7%. Both calculations are mathematically possible, but they answer different questions. Label the first as a first-ten-days comparison rather than giving both the same month-on-month label. These figures illustrate the rule; they are not customer results.
Equivalent progress does not always mean equal calendar days
A business operating mainly on weekdays may prefer the same number of working days. A seasonal campaign may need a campaign-relative window instead. Ask the business owner to choose calendar days, working days or campaign progress, and keep the calendar version with the definition. Leap years, weeks crossing month boundaries and shorter months require a documented rule. Do not silently drop dates that lack an obvious counterpart.
| Condition | Suggested display | Evidence to check |
|---|---|---|
| Current period still open | Equivalent-progress comparison with a cutoff | The actual date sets for both periods |
| Baseline is zero | Current amount and a zero-baseline note | Zero is genuine, not missing data |
| Baseline is negative | Original amounts and absolute change | An approved business interpretation |
| One period is incomplete | Not comparable yet, with the gap stated | Source completeness status |
Do not hide exceptional baselines with formatting
The common growth formula subtracts the baseline from the current value and divides by the baseline. Division by zero does not provide an ordinary growth rate. Microsoft's DAX DIVIDE documentation says that a zero denominator returns an alternative constant, or BLANK by default. This describes a tool behavior. Replacing an undefined result with zero does not establish that business activity was unchanged.
Suppose profit moves from minus 100 to minus 50. The conventional formula returns minus 50%, although the loss has narrowed. A clearer business statement may be “loss reduced by 50,” together with both original values. If an absolute-baseline formula is used, name and document it separately. Also test transitions from profit to loss and from loss to profit. Their interpretation should not depend on whichever arrow happens to appear.
Accept the calculation with reproducible samples
For each boundary case, retain both periods' detail records, filters, cutoffs, formula and expected wording. Test a month boundary, genuine zero, missing data, negative values, late records and a change of organizational scope. When users change a filter, the period label must change with the number. Include the same conditions in exports so that a screenshot of a percentage is not the only surviving evidence.
For data visualization development services, these samples can become the acceptance attachment for comparison metrics. How to Build a Metric Definition Table for an Executive Dashboard covers the underlying definition fields. If the detail records already disagree, use Why Do Data Dashboard, ERP and Excel Figures Differ, and How Can You Find the Cause? to investigate the source difference before deciding how to visualize the comparison.
What happens when the month is confirmed?
A month-to-date result can be marked provisional and followed by a complete-month result after the agreed close. A confirmed version used in a meeting should not be overwritten without explanation. If users need both equivalent-progress and complete-month comparisons, expose two clearly named views. Do not let one trend line silently change its meaning at month-end. The decision to recalculate after late postings belongs to the report-revision process and needs a named owner.