Forecast accuracy and bias

Key takeaway: Accuracy and bias tell you how trustworthy the forecast is — and which way to nudge it.

Accuracy

How close the forecast was to what actually sold. Higher is better. Above 70% is considered healthy. Lower accuracy on a high-velocity SKU is worth investigating; lower accuracy on a slow-moving SKU is normal and less concerning.

Bias

Bias shows the direction of forecast errors. A positive bias means you consistently sold more than the forecast predicted. A negative bias means the forecast is consistently over-estimating. Consistent bias in one direction is more actionable than random errors.

What to act on

• If accuracy is consistently below 60% on a fast-moving SKU, check whether there have been unusual events (a large one-off order, a stockout period) distorting the history.
• If bias is consistently positive (you keep selling more than forecast), consider raising the service level for that SKU to build in a larger safety buffer.
• If bias is consistently negative (you keep selling less than forecast), your safety stock may be higher than needed. Reducing the service level slightly could free up cash.

Frequently asked questions

What is Forecast accuracy and bias in Forthcast?

Accuracy and bias tell you how trustworthy the forecast is — and which way to nudge it.