What it means
The most common measure is mean absolute percentage error, or MAPE, which averages the size of each period's error while ignoring whether it was too high or too low. Accuracy is then simply 100% minus the MAPE.
Ignoring the direction of errors is deliberate. A forecaster who is 20% too high one month and 20% too low the next has produced two bad forecasts, even though the errors cancel out to zero when averaged with their signs attached.
The measure earns its place because forecasts drive commitments. Inventory purchases, hiring plans, cash facilities and supplier contracts are all sized from a forecast, and an inaccurate one turns into either stockouts and missed revenue, or excess stock and tied-up cash.
Accuracy is only meaningful at a stated level and horizon. Total company revenue one month ahead is far easier to forecast than a single product line six months ahead, so a target of 95% might be undemanding for the first and impossible for the second.
The most common misuse is treating a low score as a performance failure. Some demand is genuinely volatile, so the honest response is to compare against a naive benchmark, such as simply repeating last month's actual, and ask whether the forecasting effort beats it.
In practice
Real-world examples.
Example
A consumer goods brand tracks forecast accuracy by product line and finds its top ten items score 94% while long-tail items score 61%. It moves the tail onto a simple reorder rule and stops spending planner time forecasting items nobody can predict.
Example
A staffing agency measures accuracy on its 13-week cash forecast and discovers a consistent optimism of about 7% on collections. Correcting the bias alone lifts accuracy from 88% to 95% without any change to the underlying model.
Example
A subscription software firm reports 97% accuracy on total revenue but only 72% on new bookings, because renewals are predictable and new sales are not. The board starts reviewing the two separately rather than judging the combined figure.
Think of it
“Forecast accuracy shows how good your predictions are-how close you get to actual results.
Formula
Calculation
Absolute percentage error = |Actual - Forecast| / Actual. MAPE = average of those errors across periods. Forecast accuracy = 100% - MAPE.
A distributor reviews three months of revenue forecasts.
Month 1: forecast $480,000, actual $500,000. Error = $20,000 / $500,000 = 4%.
Month 2: forecast $525,000, actual $500,000. Error = $25,000 / $500,000 = 5%.
Month 3: forecast $582,000, actual $600,000. Error = $18,000 / $600,000 = 3%.
MAPE = (4% + 5% + 3%) / 3 = 12% / 3 = 4%.
Forecast accuracy = 100% - 4% = 96%.
Notice that Month 2 was over-forecast while the other two were under-forecast. If the errors had been averaged with their signs, the result would have been (-4% + 5% - 3%) / 3 = -0.67%, suggesting a near-perfect forecast and hiding all three misses.Case study
Seen in the real world.
This illustrative and fictional case follows Thornbury Components, an invented manufacturer of industrial fittings with about 400 stock items. Its planning team reported a headline forecast accuracy of 91% and was quietly proud of it, yet the warehouse was simultaneously full of slow-moving stock and short of the parts customers actually wanted.
When the finance director broke the number down, the picture changed. Accuracy was calculated on total monthly revenue, where errors on individual items offset one another. Measured at item level, MAPE was closer to 28%, giving accuracy of about 72%, and the six largest items were being under-forecast almost every month while dozens of small items were over-forecast.
Thornbury changed what it measured rather than how it forecast. It reported accuracy separately for its top 40 items, which covered 80% of revenue, and applied a simple replenishment rule to the rest.
Within two quarters, top-item accuracy had risen to 89% and inventory had fallen by roughly $700,000. The illustrative point is that a forecast accuracy figure is only as useful as the level at which it is measured.
Watch out
Common mistakes.
- Averaging errors with their signs attached. Positive and negative misses cancel out and produce a flattering number that hides genuinely poor forecasts.
- Measuring accuracy only at total company level. Aggregation hides offsetting errors, which is precisely where the operational pain lives.
- Comparing accuracy scores between businesses with different demand patterns. A 75% score on volatile project revenue can represent better forecasting than 95% on a stable subscription base.
Questions
People also ask.
What is a good forecast accuracy?
It depends on the horizon and the level of detail, though 90% to 95% on aggregate short-term revenue and 70% to 85% at item level are common ranges.
Should I divide the error by the actual or the forecast?
Dividing by the actual is the standard MAPE convention, but whichever you choose, apply it consistently so periods remain comparable.
How do I improve accuracy quickly?
Look for systematic bias first, since a forecast that is consistently 10% high can be corrected in an afternoon, unlike random error which needs better inputs.
From the founder's library
Take it further with the book.
Build your financial confidence beyond this definition. Shihan's full-length guide, Accounting Fundamentals, takes the same plain-English approach and turns it into a complete, practical playbook for non-finance managers, business owners and students - with chapter-end quiz answers and presentation slides included.
25% off with code MMHQ25, applied at checkout. Priced in USD - checkout may show the equivalent in your local currency.
View the book and save 25%