Sales forecast accuracy: how to measure it and improve it

Sales forecast accuracy is how close your forecast for a period was to what you actually closed. The simplest measure is the absolute percentage error: the gap between forecast and actual revenue, divided by actual revenue. To improve it, fix the inputs before the tooling: clear exit criteria for each pipeline stage, close dates that are evidenced rather than hoped for, deals with more than one contact, and a weekly pipeline review that tests every committed deal. Most forecast misses start in the pipeline, not in the forecasting software.

This guide is for sales directors, CROs and founders who present a forecast to the board and want fewer surprises in the last fortnight of the quarter.

What forecast accuracy means, and how to calculate it

In Forecasting: Principles and Practice, Hyndman and Athanasopoulos define a forecast error as the difference between an observed value and its forecast. For a sales team, the observed value is closed revenue for the period and the forecast is the number you committed to.

1. Forecast error = actual − forecast. A negative number means you over-forecast.
2. Absolute percentage error (APE) = |actual − forecast| ÷ actual × 100.
3. Mean absolute percentage error (MAPE) = the average of the APEs over several periods.

You can also report accuracy as 100% minus the APE. Either way, use the same definition and the same snapshot date every period.

A worked example

A team forecasts each quarter in week two and compares the snapshot with closed revenue at quarter end (figures in £000):

The MAPE across the year is 12.8%. The more useful finding is in the error column: three quarters out of four were over-forecast, by an average of nearly £39,000 a quarter across the year. That is a pattern of optimism, not random noise, and it calls for a different fix.

Two cautions about percentage errors

• The textbook notes that percentage errors are infinite or undefined when the actual value is zero, and extreme when it is close to zero. For a new product line or a small segment, report the error in pounds instead.
• Measure at a fixed point. A forecast taken in the last week of the quarter will look accurate because most deals have already closed or slipped. Week two, or the first day of the final month, tells you more.

Why forecasts miss

These are the usual causes, and most pipelines show more than one.

• Stages without exit criteria. If “Proposal” means a seller has sent a document, and not that the buyer has agreed a decision process, the stage tells you little about the chance of closing.
• Close dates set by the seller, not the buyer. Dates drift to the end of the quarter, then into the next one.
• Single-threaded deals. One contact, often not the person who signs. When that person goes quiet or leaves, the deal stalls.
• Forecasts used as targets. Hyndman and Athanasopoulos warn that if forecasts are used as targets, they may be set low so they can be exceeded more easily, and that setting targets is different from producing forecasts. The reverse happens too, when a seller commits to the number they think the manager wants to hear.
• Stage probabilities nobody has checked. Weighted forecasts multiply each deal’s amount by a probability. In HubSpot, for example, the weighted amount is the deal amount multiplied by the deal probability, which is set per pipeline stage. If the stage probabilities do not reflect your real conversion rates, the weighted total is wrong before anyone touches it.

Exit criteria for each pipeline stage

An exit criterion is something the buyer has done or confirmed, recorded in the CRM, that must be true before a deal moves on. Write them for your own process; this is a starting point for a typical B2B sale.

Qualification frameworks such as MEDDPICC give you a shared vocabulary for these criteria. Our guide to building a sales playbook covers how to choose one and keep it practical.

A weekly pipeline review that works

The same textbook’s chapter on judgmental forecasting recommends a systematic approach: set the task clearly, document and justify forecasts, and keep records so you can compare them with what happened. It notes that asking a forecaster to document and justify their forecast leads to accountability, which can reduce bias. A weekly review is where that happens.

1. Run it from the CRM, on screen, not from a spreadsheet someone prepared the night before.
2. Start with changes since last week: deals added, deals moved into commit, close dates pushed, amounts changed.
3. Inspect committed deals one by one. For each: what has the buyer done this week, who signs, and what happens next, with a date.
4. Challenge the oldest deals. Anything that has sat in the same stage far longer than your usual cycle should be moved back or closed as lost.
5. Record the forecast in the same place every week, so you can measure accuracy later.
6. Keep it short. Thirty to forty-five minutes for a team of eight is usually enough if the CRM is up to date.

When pre-sales should be involved

Pre-sales time is limited, and when it is spent on poorly qualified deals, the forecast fills with opportunities that look advanced but are not. Agree a rule: pre-sales joins once a deal has passed qualification, with the problem, the buyer’s requirements and the decision process written down. Track which deals had pre-sales involvement and compare their conversion with those that did not. That tells you whether the rule is in the right place.

What the CRM must capture

• Stage dates, so you can see how long deals sit in each stage.
• Close date history, so you can see how often dates move.
• A next step with a date on every open deal.
• At least two contacts on deals above an agreed size.
• Forecast category set by the seller and reviewed by the manager. HubSpot describes forecast categories as a way to group deals by their likelihood to close, with categories such as Pipeline, Best case and Commit, and lets users update the category on a deal.

If the data is not there, the problem is CRM adoption, and that comes first. See how to get a sales team to keep the CRM up to date.

What a realistic target looks like for your business

There is no figure that is right for every team. A business with a few large deals a quarter will see bigger swings than one with hundreds of small ones. Measure four quarters with a fixed snapshot date, calculate your MAPE and the direction of your errors, then set a target to narrow both. Review the stage probabilities at the same time, using your own stage-to-stage conversion rates.

How we can help

Our pre-sales and deal process work defines your stages, exit criteria and the point at which pre-sales joins a deal. The CRM, Pipeline and Forecast Set-up package adds deal-strength scoring, manager dashboards and a weekly forecast routine. It takes four to six weeks, from £6,500 + VAT, with software configuration and data moves quoted separately (see pricing). Read more about our CRM and tool enablement, or book a 30-minute call.

Sources

All checked on 26 October 2026.

• 5.8 Evaluating point forecast accuracy (Hyndman and Athanasopoulos, Forecasting: Principles and Practice, 3rd edition, OTexts): forecast error, MAPE and the limits of percentage errors.
• 6.2 Key principles (same book): systematic judgmental forecasting, documenting forecasts, and keeping forecasts separate from targets.
• Set up the forecast tool (HubSpot Knowledge Base): forecast categories and weighted amount.
• Set up and manage object pipelines (HubSpot Knowledge Base): deal probability per pipeline stage.

Frequently asked questions

How do you calculate sales forecast accuracy?

Take the forecast at a fixed point in the period, then compare it with actual closed revenue. The absolute percentage error is the difference divided by the actual, times 100. Average it over several periods to get the mean absolute percentage error (MAPE).

What is a good sales forecast accuracy?

It depends on your deal size, volume and sales cycle, so a figure from another business is a poor guide. Measure your own MAPE over four quarters, note whether you usually over- or under-forecast, and set a target to narrow both.

Why is our sales forecast always too high?

Usually because deals move forward without buyer evidence, close dates are set by sellers, and stage probabilities are higher than your real conversion rates. Exit criteria and a weekly review of committed deals address all three.

Does forecasting software improve accuracy?

It can help with consistency and reporting, but it works from the data in your CRM. If stages, close dates and contacts are unreliable, the software reports the same problems faster.