Demand forecasting
Tracking Signal
By the Supply Chain Manager team · Updated October 16, 2026
A tracking signal is a running check for forecast bias: cumulative forecast error divided by mean absolute deviation, flagged when it drifts too far from zero.
When errors go both ways, the cumulative error stays near zero. When the forecast keeps missing in one direction, it climbs. Dividing by the mean absolute deviation (MAD) scales it so the same limits work across SKUs. Limits of about ±4 are a common rule of thumb.
Positive values mean demand is running above forecast, negative below. Some teams use a smoothed version that weights recent errors more heavily, so the signal responds faster to new drift and forgets old errors. Either way, the purpose is the same: to flag forecasts that need attention without reviewing every SKU by hand.
How to calculate tracking signal
Tracking Signal = Σ (Actual − Forecast) ÷ MAD MAD = mean of |Actual − Forecast|
Tracking signal example
Illustrative numbers: over six weeks the errors (actual − forecast) were +8, +5, +9, +6, +7 and +10. Cumulative error = 45, MAD = 7.5, tracking signal = 6. Demand is running consistently above forecast, so the model needs reviewing.
Why tracking signal matters for distributors
With thousands of SKUs, nobody can eyeball every forecast. A tracking signal turns 'this forecast has gone stale' into an automatic flag.
Common tracking signal mistakes
- Using limits so tight that normal noise triggers constant reviews.
- Resetting the signal without fixing the model that drifted.
Examples on this page use made-up round numbers to show the calculation. They are not benchmarks or results from real businesses.