Forecast accuracy: measuring error (MAPE, bias)
Forecast accuracy is measured with MAPE and bias: knowing your real error is the foundation of correctly sized inventory. Here are these indicators and how to read them.
By Mansour Niang, CEO & co-founder, Clevizio·Published August 28, 2026·Updated October 18, 2026·7 min readKey takeaway
Forecast accuracy is measured with two indicators: MAPE (Mean Absolute Percentage Error), the average gap between forecast and actuals as a percentage, and bias, which tells you whether you systematically over-forecast or under-forecast. A MAPE of 20% means an average gap of 20%. MAPE is misleading on low volumes.
Why measure forecast error
Every forecast is wrong; the real question is by how much and in which direction. Building a good demand forecast is a topic in its own right, which we cover elsewhere; this article looks at the next step: measuring the gap between what was forecast and what actually happened. This measurement is highly operational, since it lets you size safety stock correctly, choose between two methods, and know which SKUs you are blind on. A company that does not measure its forecast accuracy manages inventory by guesswork, overprotecting some SKUs and leaving others exposed. Knowing your error turns an uncertainty you endure into a parameter you manage.
What does MAPE mean in statistics?
MAPE (Mean Absolute Percentage Error) is the most widely used indicator. It expresses the average gap between forecast and reality as a percentage, which makes it easy to understand and to compare across SKUs. A MAPE of 20% means that, on average, the forecast deviates from actuals by 20%. Simple and intuitive, it nonetheless has limits: it behaves poorly on low volumes and intermittent demand, where a small absolute error produces a huge percentage. For those SKUs, other error measures are better suited.
| Indicator | What it measures | Watch out |
|---|---|---|
| MAPE | Average error as a percentage | Misleading on low volumes |
| Bias | Tendency to over- or under-forecast | Must be tracked |
| MAD | Mean absolute error in units | Less meaningful in relative terms |
What is forecast bias?
People often focus on the size of the error and forget its direction. Bias measures whether you tend to systematically over-forecast or under-forecast. This is critical, because a biased error is far more dangerous than a symmetrical one: constantly over-forecasting inflates inventory, while constantly under-forecasting multiplies stockouts. A forecast can show a decent MAPE while being heavily biased, which quietly destroys value. Tracking bias alongside MAPE gives a complete picture of forecast quality.
How to use these measures
These indicators are calculated by SKU and by product family, over a rolling window, to spot where the forecast is reliable and where it is not. Where error is high, you increase safety stock or change method. Where bias is pronounced, you correct the model or the process. Reaching zero error is impossible; the realistic goal is to know your error and act on it. That clear-sightedness is what separates mature management from flying blind.
Reducing error is not enough
Measuring and reducing forecast error is essential, but there is a ceiling: even the best forecast is wrong about events it did not see coming. Gaining a few points of MAPE does not protect you from a supplier failure or a sudden spike, which are unpredictable by nature. Forecast accuracy and the quality of the decision when the forecast is wrong are two distinct and complementary topics. You work on the first to reduce avoidable errors, and you prepare the second to absorb the inevitable ones.
This article starts from a conviction: the perfect forecast doesn't exist, what matters is the decision that follows.
- Why measure forecast error
- What does MAPE mean in statistics?
- What is forecast bias?
- How to use these measures
- Reducing error is not enough
Clevizio
Clevizio is the decision layer that plugs into your existing tools: it detects disruptions, simulates your options and keeps a record of every trade-off. See how the decision copilot works.
Mansour Niang
CEO & co-founder, Clevizio
A question about your disruptions? Book 30 minutes with me to estimate what they really cost you, with no commitment.
Frequently asked questions
What is MAPE in forecasting?
MAPE (Mean Absolute Percentage Error) is the average gap between forecast and reality, expressed as a percentage. A MAPE of 20% means the forecast deviates from actuals by 20% on average. It is intuitive but misleading on low volumes.
Why track forecast bias?
Because a biased error is more dangerous than a symmetrical one. Systematically over-forecasting inflates inventory, while under-forecasting multiplies stockouts. A forecast can have a decent MAPE while being heavily biased, which quietly destroys value.
Can forecast error ever reach zero?
No. No forecast is 100% accurate. The goal is not zero error but knowing your error, so you can size inventory and prepare decisions for the inevitable deviations.
Finally put a number on what disruptions cost you every year.
Let's start by estimating what your disruptions really cost you. It's free, it takes 30 minutes, and you leave with a figure nobody in your company knows today.