Editorial team ·

Compare a forecast with the result without changing the measure

Editorial cover: Compare a forecast with the result without changing the measure
AI-generated illustration.

The forecast said 1,000 visits in the first month. The later report says 800 visits, but it covers the first three weeks. Before writing that the target was missed, check whether both numbers measure the same period. Then find out whether the forecast was a target or a prediction.

Match the unit and time window

Three values in an invented forecast comparison
Source Period Value
Forecast Four weeks 1,000 visits
Interim report Three weeks 800 visits
Comparable final report Same four weeks 950 visits

In the fictional case, the first public forecast covers four weeks. The interim report covers three. A later month-end release counts 950 visits under the same rule. Compare the forecast with that release if the area and coverage still match.

The difference is 50 visits below the forecast in this simplified example. That describes an observed difference. You would need further evidence to explain why it occurred.

Review whether the forecast gave a range or more than one scenario. A central estimate may differ from the result even when the result sits within the stated range. Preserve those limits if they matter.

If the report changed the count definition, say so before comparing. For the reader, a clear reason two numbers differ in scope is more useful than an exact percentage with the wrong base.

Keep the source’s assumptions with the comparison

  1. Match the figures
    • Compare time, area and counting rule.
    • Check whether the source calls it a forecast or target.
  2. Report the observed difference
    • Use the comparable result only.
    • Leave the cause open unless supported.

Questions and answers

Does 950 prove the forecast was bad?
Check the stated range, assumptions and conditions before assessing the forecast.
Can I compare 800 directly with 1,000?
The example uses different time windows. Keep the figures separate unless you find a valid common basis.
Should I mention the percentage difference?
If useful, state the base. Fifty fewer than a forecast of 1,000 is five percent below it.