Forecast accuracy formula for sales teams
The forecast accuracy formula for one period is accuracy = 1 - |actual - forecast| / actual. Over several periods, use weighted accuracy (1 minus the sum of absolute errors divided by the sum of actuals), and look at MAPE and bias next to it: accuracy says how close you were, bias says in which direction you keep missing.
Forecast accuracy calculator
Bias: +18.1% (over-forecast)
Month 1: 82.9%, Month 2: 96.2%, Month 3: 52.5%
Accuracy per period = 1 - |actual - forecast| / actual. Weighted accuracy = 1 - sum of absolute errors / sum of actuals.
The four formulas
| Metric | Formula | What it tells you |
|---|---|---|
| Accuracy, one period | 1 - |A - F| / A | How close one forecast was |
| Weighted accuracy | 1 - Σ|A - F| / ΣA | Accuracy over several periods, big periods weigh more |
| MAPE | average of |A - F| / A | The typical size of a miss, every period weighs the same |
| Bias | Σ(F - A) / ΣA | Positive: you over-forecast; negative: you under-forecast |
A is the actual booked revenue, F the forecast you committed to at a fixed point (for example, the first Monday of the month). Fix that point and keep it: an accuracy measured on the forecast of the last day is not a forecast.
Worked example
The calculator's default, three months:
| Month | Forecast | Actual | Error | Accuracy |
|---|---|---|---|---|
| 1 | $820,000 | $700,000 | $120,000 over | 82.9% |
| 2 | $760,000 | $790,000 | $30,000 under | 96.2% |
| 3 | $900,000 | $610,000 | $290,000 over | 52.5% |
| Total | $2,480,000 | $2,100,000 | $440,000 absolute | 79% weighted |
MAPE is 22.8%: a typical month misses by about a fifth. Bias is +18.1%: over the quarter, the team committed 18% more than it booked. The two months of over-forecast are not noise; they are a pattern.
When the metrics disagree
- High accuracy, high bias. Misses are small but always in the same direction: the forecast is systematically optimistic or sandbagged. Fix the process, not the math.
- Good weighted accuracy, bad MAPE. Large months are right and small months are wrong. The quarter looks fine, monthly cash planning does not.
- Accuracy falls while coverage rises. More pipeline, fewer real deals: check the share of deals with no recent activity, and the coverage on the clean pipeline with the pipeline coverage calculator.
Measure Commit separately
The total forecast hides where the error comes from. Split it by forecast category and measure the Commit line on its own:
Commit hit rate = amount won from deals that were in Commit at the snapshot / amount in Commit at the snapshot
| Team | Commit at snapshot | Won from that Commit | Commit hit rate | Reading |
|---|---|---|---|---|
| A | $600,000 | $540,000 | 90% | Commit means commit |
| B | $600,000 | $330,000 | 55% | Commit is a hope category |
Team B can still post a decent total accuracy if Best Case deals close by surprise, but the number it committed to was wrong. A low Commit hit rate usually traces back to deals that stayed in Commit without buyer activity.
Pick a snapshot point and keep it
Accuracy is only comparable if the forecast is taken at the same moment every period:
- Monthly forecast: the first business day of the month, or the end of week one.
- Quarterly forecast: the start of week three of the quarter is common, early enough to act and late enough to be informed.
- Never the last day. A forecast taken when most deals are already closed measures bookkeeping, not forecasting.
Write the snapshot rule next to the metric, so the number means the same thing next year.
Related error measures
| Name | Relation to this page |
|---|---|
| WAPE (weighted absolute percentage error) | Equals 1 minus weighted accuracy: 21% on the worked example |
| MAPE (mean absolute percentage error) | The average of per-period errors, 22.8% on the example |
| Forecast bias, or mean percentage error | The signed version: +18.1% on the example |
| Hit rate, or attainment of forecast | Actual / forecast, 84.7% on the example's total |
How sales teams improve it
- Take silent deals out of Commit. A deal with no activity for more than 14 days is a hope, not a commitment.
- Ask a reason for every pushed close date. Pipeline Inspection highlights the pushes; the review asks why.
- Forecast on the clean pipeline. Compare the reps' call with the pipeline that excludes zombie deals.
- Measure every period, the same way. Same formula, same snapshot point, same segment.
- Check the inputs. A falling win rate under a flat forecast is the earliest warning.
Accuracy by manager
| Manager | Forecast | Actual | Accuracy | Bias |
|---|---|---|---|---|
| North | $300,000 | $280,000 | 92.9% | +7.1% |
| South | $250,000 | $160,000 | 43.8% | +56.3% |
| West | $200,000 | $210,000 | 95.2% | -4.8% |
| Team | $750,000 | $650,000 | 81.5% weighted | +15.4% |
Weighted accuracy for the team is 1 - (20,000 + 90,000 + 10,000) / 650,000 = 81.5%. That respectable figure hides one manager whose call missed by more than half. That is where the forecast review should start.
Report it in one line
"Weighted accuracy 79%, MAPE 22.8%, bias +18.1% over the last three months, snapshot on the first business day." One line, three numbers and the snapshot rule: enough for anyone to compare next quarter.
Measure it in Salesforce
- Snapshot the forecast. On the snapshot date, export the forecast by manager (Forecasts tab) or save a report of open opportunities by forecast category.
- Wait for the period to close. Then run a Closed Won report for the same people and period.
- Compute per manager and in total with the formulas above, or paste the figures into the calculator.
- Keep a log. One row per period: forecast, actual, accuracy, bias. Four periods are the minimum to read a trend.
Common mistakes when measuring it
- Measuring against the latest forecast. The forecast changes until the last day; measure against the snapshot.
- Averaging percentages across teams of different size. Use weighted accuracy for the total, MAPE to compare periods.
- Ignoring the sign. Two months at +20% and -20% average to zero bias and hide two bad calls; read accuracy and bias together.
- Changing the formula when the number is bad. Pick one formula, write it down, and keep it for at least four periods.
- Measuring only the team. Measure each manager's call too: that is where coaching starts.
How common is a poor forecast?
Common enough to be the norm. In a 2020 Gartner survey, fewer than half of sales leaders and sellers, 45%, had high confidence in their organization's forecast accuracy. A later Gong press release put the share of sales leaders confident in their team's revenue forecast at 24%. Both are surveys of confidence, not measured accuracy, which is why measuring your own is the first step.
Start with four periods of history computed the same way, then decide what to change. For tools that produce or check the forecast, see sales forecasting software.
Frequently asked questions
What is the forecast accuracy formula?
Forecast accuracy for one period is 1 minus the absolute difference between actual and forecast, divided by actual. A $760,000 forecast against $790,000 actual is 1 - 30,000 / 790,000 = 96.2%.
What is the difference between accuracy, MAPE and bias?
Accuracy says how close the forecast was, MAPE averages the size of the errors across periods, and bias says in which direction you usually miss. A forecast can be accurate on average and biased every month.
What is a good sales forecast accuracy?
Benchmarks are rare and mostly vendor-published. In a 2020 Gartner survey, only 45% of sales leaders and sellers had high confidence in their organization's forecast accuracy, so measure your own trend first.
How can I improve sales forecast accuracy?
Remove deals with no recent activity from Commit, require a reason for every pushed close date, and compare the forecast with the clean pipeline each week. Then measure accuracy every period with the same formula.
Related guides
Sources
- Forecast accuracy, Institute of Business Forecasting, read 21 Oct 2026
- How do I measure forecast accuracy, Forecast Pro, Aug 2020
- Gartner says less than 50% of sales leaders and sellers have high confidence in forecasting accuracy, Gartner, 12 Feb 2020
- Only 24% of sales leaders are confident in their team's revenue forecast, Gong press release, read 21 Oct 2026
TruthPipe is an independent product. It is not affiliated with, sponsored or endorsed by Salesforce or Anthropic. Salesforce, Agentforce and Claude are trademarks of their owners.