Sales forecast accuracy tells you how close your forecast came to the revenue you actually closed. The standard formula is:
Forecast accuracy = (1 − |Actual − Forecast| ÷ Actual) × 100
If your team committed $1,000,000 for the quarter and closed $900,000, accuracy is (1 − 100,000 ÷ 900,000) × 100 = 88.9%.
That single number is a start, but it hides most of what a sales leader needs to know: which reps are consistently optimistic, whether "Commit" really means commit, and how many deals quietly slip into next quarter. This guide covers the formulas, the metrics worth tracking, worked examples in revenue (not units), and how to set an accuracy target from your own data.
What is sales forecast accuracy?
Sales forecast accuracy (also called sales forecasting accuracy) measures the gap between the revenue a sales team predicted for a period and the revenue it actually closed in that period. It is usually reported as a percentage, where 100% means the forecast matched actual bookings exactly.
Two things make sales forecasting different from demand or inventory forecasting:
- The unit is revenue from a small number of deals. One $200K deal slipping can move a quarter more than fifty small ones. That changes which metrics are useful.
- The forecast is partly human judgment. Reps and managers call deals into categories like Commit and Best Case. That means accuracy problems are often behavior problems (optimism, sandbagging, stale close dates), not model problems.
If you forecast units for inventory or supply-chain planning, the metrics below still apply, but the examples and benchmarks in this guide are written for B2B revenue teams. For a statistical treatment of forecast error, Hyndman and Athanasopoulos' Forecasting: Principles and Practice is the standard free reference.
How to calculate sales forecast accuracy
Use the forecast you submitted at a fixed point (for example, the start of the quarter or week 6) and compare it with closed-won revenue at the end of the period. The sales forecast accuracy formula most teams use is:
Forecast accuracy (%) = (1 − |Actual − Forecast| ÷ Actual) × 100
Forecast error (%) = |Actual − Forecast| ÷ Actual × 100
Worked example:
Value | |
|---|---|
Quarter forecast (submitted week 1) | $1,000,000 |
Closed-won revenue | $900,000 |
Absolute error | $100,000 |
Forecast error | 100,000 ÷ 900,000 = 11.1% |
Forecast accuracy | 100% − 11.1% = 88.9% |
Two practical notes:
- Divide by actual, not forecast. Some teams use Actual ÷ Forecast (here, 90%). That is a "forecast attainment" ratio, not an error measure, and it treats over- and under-forecasting differently. Pick one definition and keep it.
- Accuracy can go below zero. If a rep forecasts $100K and closes $40K, the error is 150% and the formula returns −50%. Most teams floor accuracy at 0% for reporting, but keep the raw error for analysis.
Forecast accuracy metrics: formulas and when to use each
A single accuracy percentage is fine for a board slide. To manage the forecast you need a few more metrics. In the formulas below, A is actual revenue, F is forecast revenue, n is the number of items (reps, segments or periods), and Σ means "sum across all items."
Metric | Formula | What it tells you | Use it when |
|---|---|---|---|
Forecast accuracy % | (1 − |A − F| ÷ A) × 100 | How close one forecast was | Reporting a single period at team or company level |
MAPE (mean absolute percentage error) | (1/n) × Σ (|A − F| ÷ A) × 100 | Average % error, every item weighted equally | Items are similar in size and none are close to zero |
WAPE (weighted absolute percentage error) | Σ |A − F| ÷ Σ A × 100 | Total error as a share of total revenue | Reps or segments vary a lot in size (most B2B teams) |
WMAPE (weighted MAPE) | Σ (wᵢ × |Aᵢ − Fᵢ| ÷ Aᵢ) ÷ Σ wᵢ × 100 | % error weighted by what matters to you (margin, priority) | Some revenue matters more than other revenue |
MAE / MAD (mean absolute error / deviation) | (1/n) × Σ |A − F| | Average miss in dollars | You want the error in money, not percentages |
RMSE (root mean squared error) | √[(1/n) × Σ (A − F)²] | Average miss, with big misses penalized more | One large miss hurts more than several small ones |
sMAPE (symmetric MAPE) | (1/n) × Σ (|A − F| ÷ ((A + F) ÷ 2)) × 100 | % error that treats over- and under-forecasts more evenly | Actuals are sometimes very small |
Forecast bias | Σ (F − A) ÷ Σ A × 100 | Whether you consistently forecast too high (+) or too low (−) | Always. Bias is the metric that drives coaching |
Worked example: one quarter, five reps
Here is a team that forecast $1,000,000 and closed $850,000 (figures in $K):
Rep | Forecast | Actual | Error (F − A) | Absolute % error | Bias |
|---|---|---|---|---|---|
Rep A | 300 | 240 | +60 | 25.0% | +25.0% |
Rep B | 250 | 265 | −15 | 5.7% | −5.7% |
Rep C | 200 | 150 | +50 | 33.3% | +33.3% |
Rep D | 150 | 155 | −5 | 3.2% | −3.2% |
Rep E | 100 | 40 | +60 | 150.0% | +150.0% |
Team | 1,000 | 850 | +150 | +17.6% |
The same quarter, scored five ways:
Metric | Calculation | Result |
|---|---|---|
Team forecast accuracy | 1 − 150 ÷ 850 | 82.4% |
MAPE | (25.0 + 5.7 + 33.3 + 3.2 + 150.0) ÷ 5 | 43.4% |
WAPE | (60 + 15 + 50 + 5 + 60) ÷ 850 | 22.4% |
MAE | 190 ÷ 5 | $38K |
RMSE | √[(60² + 15² + 50² + 5² + 60²) ÷ 5] | $44.6K |
Bias | +150 ÷ 850 | +17.6% |
What this shows:
- Team accuracy (82.4%) looks better than WAPE implies (1 − 22.4% = 77.6%). Team accuracy nets errors against each other, so Rep B's under-forecast partly cancels Rep A's over-forecast. WAPE adds up every miss.
- MAPE (43.4%) is distorted by the smallest rep. Rep E's $60K miss on a $40K actual is a 150% error, and MAPE gives it the same weight as Rep A's $240K book. This is why MAPE is a poor headline metric for most sales teams.
- Bias (+17.6%) is the real story. Three of five reps over-forecast. The fix is a coaching conversation, not a new model.
WAPE vs. WMAPE: what's the difference?
The two names are often used interchangeably, but they are not the same:
- WAPE divides total absolute error by total actual revenue: Σ |A − F| ÷ Σ A. Larger reps or segments automatically count more, because their errors are bigger in dollars.
- WMAPE lets you choose the weights (wᵢ). If you weight each item by its actual revenue (wᵢ = Aᵢ), WMAPE gives exactly the same result as WAPE. If you weight by something else, such as gross margin or strategic priority, the result changes.
Using the five-rep example above, with WMAPE weighted by gross margin (Reps A and B sell an 80%-margin product, C and D 50%, E 30%):
Metric | Weighting | Result |
|---|---|---|
WAPE | Actual revenue (built in) | 22.4% |
WMAPE, wᵢ = actual revenue | Actual revenue | 22.4% (identical to WAPE) |
WMAPE, wᵢ = actual × margin | Gross-profit dollars | 18.6% |
The margin-weighted error is lower because the biggest percentage miss (Rep E) sits in the lowest-margin product. If leadership plans around gross profit rather than bookings, that is the more useful number.
Rule of thumb: use WAPE as your default headline metric. Use WMAPE only when you can name and defend the weight.
MAPE vs. WAPE: which should you report?
Situation | Use | Why |
|---|---|---|
Reps, segments or regions vary in size | WAPE | Weights errors by revenue, so small books can't dominate |
Any item can have very small or zero actuals | WAPE or sMAPE | MAPE divides by actual and becomes huge or undefined |
Comparing accuracy across companies or teams of different size | MAPE or WAPE (state which) | Both are percentages, but say which one you used |
Coaching an individual rep across quarters | Bias + absolute % error per quarter | Shows direction and size of the miss |
Accuracy by forecast category: Commit, Best Case, Pipeline
Most B2B teams split the forecast into categories. Measuring each category separately tells you whether your category rules work.
Category close rate = Closed-won from category ÷ Value in category at the snapshot date
Continuing the example quarter ($K):
Category | Value at week 1 | Closed-won by quarter end | Category close rate |
|---|---|---|---|
Commit | 600 | 570 | 95.0% |
Best Case | 250 | 130 | 52.0% |
Pipeline | 900 | 150 | 16.7% |
Total closed | 850 |
How to read it:
- Commit should close at a very high rate. If Commit regularly closes well below what you expect, your entry criteria for Commit are too loose (for example, deals without a confirmed decision date or signer).
- Best Case is where most forecast error hides. Track its close rate over four or more quarters and use that historical rate, rather than a gut call, when you add Best Case to the forecast.
- Pipeline close rate tells you how much in-quarter creation and acceleration you can realistically count on.
Accuracy by rep: optimism vs. sandbagging
Team-level accuracy can look fine while two reps miss in opposite directions. Track bias per rep over several quarters:
Quarter | Rep A forecast | Rep A actual | Rep A bias | Rep D forecast | Rep D actual | Rep D bias |
|---|---|---|---|---|---|---|
Q1 | 300 | 240 | +25.0% | 120 | 150 | −20.0% |
Q2 | 280 | 230 | +21.7% | 130 | 160 | −18.8% |
Q3 | 260 | 215 | +20.9% | 110 | 140 | −21.4% |
Q4 | 300 | 250 | +20.0% | 125 | 150 | −16.7% |
Year | 1,140 | 935 | +21.9% | 485 | 600 | −19.2% |
- Rep A is consistently optimistic (about +22%). Deals are called too early or at inflated values.
- Rep D is consistently sandbagging (about −19%). Deals are held back to beat the number.
Because both patterns are consistent, you can correct for them while you coach. Since bias = (F − A) ÷ A, the expected actual is F ÷ (1 + bias):
- Rep A forecasts $300K next quarter → expected close ≈ 300 ÷ 1.219 = $246K
- Rep D forecasts $120K → expected close ≈ 120 ÷ 0.808 = $148.5K
Use this as a manager's adjustment, not a replacement for fixing the behavior. The goal is for each rep's bias to trend toward zero.
Deal slippage: the hidden driver of forecast error
A deal can be won and still wreck this quarter's forecast if it closes next quarter. Track slippage separately from win rate:
Slip rate = Deals forecast to close this period whose close date moved out ÷ All deals forecast to close this period
Example: 40 deals were forecast to close this quarter at week 1. By quarter end, 14 had their close date pushed into a later quarter. Slip rate = 14 ÷ 40 = 35%.
What to track alongside it:
- Slipped value, not just count. Five small slips matter less than one large one.
- Repeat slippers. Deals whose close date has moved two or more times are the strongest candidates to drop out of Commit.
- Slip reasons. Legal review, budget approval and a missing decision-maker are fixable with earlier qualification. "No reason" usually means the close date was never real.
For more on keeping pipeline data clean enough to forecast from, see our guide to pipeline health metrics.
Measure at a fixed point: forecast horizon
The same quarter can look accurate or inaccurate depending on when you take the forecast. A forecast submitted the week before quarter end will almost always look accurate. One submitted in week 1 is the real test.
Pick fixed snapshot points and measure accuracy at each:
Snapshot | What accuracy here tells you |
|---|---|
Week 1 of the quarter | Whether your planning and hiring decisions are built on a reliable number |
Mid-quarter (around week 6) | Whether your mid-quarter call to finance is trustworthy |
Two weeks before quarter end | Whether late-quarter deal management is under control |
Report the week-1 number to leadership. It is the one that drives planning, and it improves only when the underlying process improves.
What is a good sales forecast accuracy rate?
There is no single verified industry benchmark for "good" B2B forecast accuracy. Published figures vary by company size, deal size, sales cycle and how accuracy is defined, and many widely shared numbers have no traceable source. Here is what primary research does show:
Finding | Source |
|---|---|
Only 45% of sales leaders and sellers have high confidence in their organization's forecasting accuracy | Gartner, State of Sales Operations Survey (press release, Feb 2020) |
Only 47% of respondents believe their organization has high-quality data | |
4 in 5 sales and finance leaders missed a quarterly sales forecast in the past year; over half missed it two or more times | Xactly, 2024 Sales Forecasting Benchmark Report (400 respondents, Jul 2024) |
Reporting systems that can't access historical CRM or performance data are the most common roadblock to accurate forecasting (66%) |
In short, missing the forecast is the norm, not the exception, and data quality is the most cited cause.
How to set your own accuracy target
Because outside benchmarks are inconsistent, the most defensible target comes from your own history:
- Calculate week-1 WAPE and bias for each of your last four quarters. This is your baseline.
- Set the target as an improvement on that baseline, for example cutting WAPE by a few points per quarter, rather than adopting an outside number.
- Add a bias target of close to zero at team level and for every rep.
- Track Commit close rate and slip rate as the two leading indicators. If either moves the wrong way, accuracy will follow.
How to improve sales forecast accuracy
Each fix below targets a specific metric, so you can see whether it worked.
Problem you see | Likely cause | What to do | Metric that should move |
|---|---|---|---|
High positive bias for one or more reps | Optimism, deals called too early | Review those reps' Commit deals line by line; require a confirmed decision date and signer for Commit | Rep bias toward 0% |
Negative bias for a rep | Sandbagging | Compare their Best Case close rate with the team's; adjust their forecast by historical bias while coaching | Rep bias toward 0% |
Commit closes far below expectations | Loose category rules | Write down entry criteria for each category and audit them in the weekly call | Commit close rate |
High slip rate | Unrealistic close dates | Flag deals whose close date has moved twice; require a reason code for each change | Slip rate, slipped value |
Week-1 accuracy much worse than late-quarter accuracy | Forecast is built late, not managed early | Lock and save a week-1 snapshot every quarter and review it at quarter end | Week-1 WAPE |
Accuracy swings with no clear pattern | Missing or stale deal data | Make close date, amount and stage required fields; clean stale deals before the snapshot | WAPE, variance between quarters |
If you are evaluating AI-based forecasting, our article on AI sales forecasting accuracy covers what to test before you trust it. For the forecasting methods themselves, see sales forecasting methods and pipeline forecasting.
Tracking forecast accuracy in your CRM
Every metric in this guide depends on three fields being reliable for every deal: amount, close date and stage or category, plus a saved snapshot of the forecast at a fixed date. If those live in spreadsheets and inboxes, you can't calculate bias by rep or slip rate without manual work.
Gain.io keeps deals, stages, values and activity in one sales pipeline, so the inputs for forecast accuracy are in one place and up to date. For the pipeline numbers that feed the forecast, see our guide to pipeline metrics.


