How to calculate forecast accuracy
Forecast accuracy measures how close your predictions come to reality. It is the metric that makes a forecast trustworthy: a consistently accurate forecast lets leadership plan hiring and spend, while a wild one erodes credibility every quarter.
Short answer
Calculate forecast accuracy by comparing what you forecast to what actually closed, usually as one minus the absolute error divided by actual. For example, forecasting 1 million dollars when 950,000 closed is 95 percent accuracy. Track it by rep and forecast category over time to find who and what is consistently off.
Step by step
Record the forecast at a fixed point
Capture the committed forecast at a consistent moment each period, such as the start of the quarter, so you are always comparing the same snapshot.
Compare to actual results
At period end, compare forecast to what actually closed and compute the error.
- Accuracy = (1 - |forecast - actual| / actual) x 100
- Snapshot the forecast at a consistent time
- Track both over- and under-forecasting
Track direction, not just magnitude
Note whether the rep or team consistently over- or under-forecasts. Directional bias is coachable in a way that random error is not.
Segment by rep and category
Break accuracy down by rep and forecast category (commit, best case) to find where the number is unreliable.
Feed it back into the process
Use the pattern to recalibrate: sandbaggers and happy-ears reps each need different coaching to tighten accuracy.
How Ardovo helps
Ardovo snapshots each forecast and scores accuracy against actuals by rep and category automatically, so Rook can show you who consistently sandbags or over-commits and by how much, turning forecast reviews into targeted coaching.
Frequently asked questions
What is a good forecast accuracy?
Mature teams often target 90 percent or better on the committed forecast within a period. What matters most is consistency and low bias: a forecast that is reliably close, in either direction, is far more useful than one that swings unpredictably.
Should I measure over-forecasting and under-forecasting differently?
Track both the magnitude and the direction of error. Persistent over-forecasting (happy ears) and persistent under-forecasting (sandbagging) are different coachable behaviors. Directional bias is more fixable than random error.
When should I snapshot the forecast?
At a consistent point each period, commonly the start of the quarter or a set weekly cadence. Comparing the same snapshot to actuals every time keeps the accuracy metric fair and reveals genuine trends rather than timing noise.