Forecast Accuracy and MAPE: A Practical Primer

Forecast accuracy drives inventory levels, service rates, and customer experience. Most retailers track it loosely. Best-in-class operators measure it religiously at SKU-week level and improve it as a permanent program.

MAPE

Mean Absolute Percentage Error — the most common forecast accuracy metric. MAPE = average of |Forecast − Actual| / Actual. Best-in-class is 15–25 percent (i.e., 75–85 percent accuracy) at SKU-week level.

WAPE

Weighted Absolute Percentage Error normalizes for SKU volume. WAPE = sum(|F−A|) / sum(A). Less skewed by low-volume SKUs than MAPE. Use WAPE for aggregate reporting and MAPE for SKU-level diagnostics.

Bias

Bias = average (Forecast − Actual). Tells you whether your forecast is systematically high or low. Often more actionable than absolute error — bias usually has a process root cause.

Improving accuracy

Clean historical data, add external signals (weather, promotions), use ML models for high-volume SKUs, and review forecast vs. actual weekly. Each lever adds 2–5 points of accuracy.