Forecast Accuracy Metrics: MAPE, Bias, and Beyond

Measuring forecast accuracy is essential for improving your demand planning process. But using the wrong metric — or only one metric — can hide important problems. Here's a comprehensive guide to the metrics that matter.

MAPE: Mean Absolute Percentage Error

MAPE = (1/n) × Σ |Actual - Forecast| / Actual × 100

MAPE is the most common accuracy metric. It's intuitive (expressed as a percentage) and easy to communicate. But it has limitations:

• Undefined for zero actuals (division by zero)
• Asymmetric (over-forecasts and under-forecasts penalized differently)
• Misleading for low-volume items (a 1-unit error on a 2-unit actual is 50% MAPE)

Weighted MAPE

Weighted MAPE addresses the low-volume problem by weighting errors by volume or revenue:

WMAPE = Σ |Actual - Forecast| / Σ Actual × 100

This gives a more business-relevant view of accuracy because high-volume items contribute proportionally more.

Forecast Bias

Bias = Σ (Forecast - Actual) / Σ Actual × 100

Bias tells you the direction of your errors. A positive bias means you're systematically over-forecasting; negative means under-forecasting. Unlike accuracy metrics, bias is signed — and that's the point.

Forecast Value Add (FVA)

FVA measures whether each step in your planning process improves the forecast:

FVA = Accuracy(Step N) - Accuracy(Step N-1)

If a planner's overrides make the forecast less accurate than the statistical baseline, that's negative FVA — a clear signal for process improvement.

Choosing the Right Metrics

| Metric | Best For | Watch Out For | |--------|----------|---------------| | MAPE | General accuracy communication | Low-volume items, zero actuals | | WMAPE | Business-weighted accuracy | Can hide problems in the tail | | Bias | Systematic error detection | Doesn't measure magnitude | | FVA | Process improvement | Requires step-by-step tracking |

Key Takeaways

1. Use multiple metrics — no single metric tells the whole story
2. Segment your accuracy reporting by volume, category, and newness
3. Track bias separately from accuracy to detect systematic issues
4. Implement FVA to ensure every planning step adds value