Control · C85
Measure forecast accuracy and bias
Does plan variance come from normal volatility or systematic forecasting bias?
Direct answer
The analysis combines weighted absolute error, signed bias and stability by horizon to distinguish imprecision from recurring optimism.
01Keep the method, correct the bias, shorten the horizon or change the assumptions.
02The rule, the evidence and the next review date stay visible.
How to use this
Move from the question to the deliverable.
Does plan variance come from normal volatility or systematic forecasting bias?
Keep the method, correct the bias, shorten the horizon or change the assumptions.
Original forecast-accuracy series
Original chart · SQ-85
Tell an imprecise forecast from a systematically optimistic one
An average error close to zero can hide large errors that offset each other. Precision requires an absolute measure; bias keeps the sign of the variances.
Fictitious series of marketing pipeline planned and delivered over six months, at constant scope.
- 01Freeze
Keep the forecast published at each horizon instead of replacing it with the latest estimate.
- 02Measure
Calculate WAPE for precision and signed bias for optimism or caution.
- 03Segment
Compare by horizon, channel and type of initiative before aggregating.
- 04Correct
Change the method or the assumptions when the bias persists for three periods.
Six months of forecast: 9.7% error and optimistic bias
Five months out of six are overestimated. The error is therefore not only volatility: the method carries a bias.
WAPE = Σ|actual − forecast| ÷ Σ actual; bias = Σ(actual − forecast) ÷ Σ actual| Month | Forecast | Actual | Absolute error | Bias | Reading |
|---|---|---|---|---|---|
| M1 | €420k | €390k | €30k | −€30k | Optimistic |
| M2 | €460k | €402k | €58k | −€58k | Optimistic |
| M3 | €480k | €455k | €25k | −€25k | Optimistic |
| M4 | €500k | €512k | €12k | +€12k | Cautious |
| M5 | €535k | €470k | €65k | −€65k | Optimistic |
| M6 | €560k | €486k | €74k | −€74k | Optimistic |
01WAPE reaches 9.7% while the signed bias reaches −8.8%: the forecast is affected above all by systematic optimism.
02The bias persists over five months and justifies correcting the planned conversion rates.
03The target is not rewritten: a new forecast is kept alongside the approved plan.
Fictitious data. WAPE becomes unstable when the delivered volumes are very low or nil.
Coherence controls
Minimum conditions before validation.
The status, the evidence and the owner must be available for review.
The status, the evidence and the owner must be available for review.
The status, the evidence and the owner must be available for review.
The status, the evidence and the owner must be available for review.
Usable deliverable
Original forecast-accuracy series
Explore the decision further with an interactive control, or start from a structured template to produce your deliverable.