How do you calculate sales forecast accuracy?
The common formula for one period is: forecast accuracy = 1 − (|actual − forecast| ÷ actual). A forecast of €3M against €1M closed misses by twice the amount that closed, so the accuracy is below zero.
Two choices decide whether the number means anything:
- Fix the moment. Measure the forecast as it stood at a set point, for example the start of the quarter or week six, not the last call before quarter end. Otherwise you measure how fast the forecast was corrected, not how good it was.
- Measure deals as well as totals. A total can come out right because one deal slipped and another closed by surprise. Count how many of the deals forecast to close in the period actually did.
What is forecast bias?
Accuracy says how far off the forecast was. Bias says in which direction, period after period. Add up forecast minus actual over several quarters: a result above zero means the forecast leans optimistic. A team can reach acceptable accuracy in one quarter by luck and still carry a steady optimistic bias, and the bias is the part you can fix.
Why does the CRM forecast run high on large deals?
A stage‑weighted forecast multiplies each deal’s value by the weight of its stage. That works when a stage means the same thing every time. On large deals this is where it breaks:
- Stages follow activity, not proof. “Demo done” moves a deal forward even if nobody on the buyer’s side has confirmed a problem worth solving.
- One stage moves a large number. With fewer, larger deals, a single optimistic stage changes the quarter.
- Close dates follow your calendar. The buyer’s security, legal and procurement steps are often unmapped, while Forrester’s 2026 research found procurement is a decision‑maker in 53% of business buying cycles.
- The economic buyer is assumed. A deal can reach proposal without the person who releases the budget ever seeing the business case.
None of this needs anyone to be dishonest. It only needs stages to be recorded when someone believes them.
Commit, best case and pipeline: where forecast categories break
Forecast categories add judgement on top of stages: commit for deals the owner would stake their name on, best case for deals that could close, pipeline for the rest. They are useful, but without written rules they turn into labels for confidence. Give commit a definition built on evidence instead:
- The economic buyer has seen the business case and confirmed budget and timing.
- The buyer’s procurement, legal and security steps are mapped and dated.
- There is a next step with a date that both sides agreed.
A deal that misses one of these can still close. It is just not commit.
What is an earned forecast?
An earned forecast is a forecast based on verified deal evidence. Each deal counts at the stage its evidence has earned, not the stage recorded in the CRM, and the earned number is shown next to the CRM forecast. The gap between the two is the agenda for the review: which deals carry a stage they have not earned, and what evidence would close the gap.
For example, a deal recorded at Commercial Proposal whose economic buyer has not seen the business case counts at Proof Of Value in the earned forecast. Nothing is deleted; the deal is counted where the evidence puts it.
How to improve sales forecast accuracy
- Write down the evidence each stage needs, so a stage means the same thing for every deal and every owner.
- Review the largest deals every week against that evidence, not against the owner’s confidence.
- Move a deal back one stage when it has no next step that both sides agreed and dated.
- Show the earned forecast next to the CRM forecast and discuss the gap, deal by deal.
- Measure accuracy at a fixed point and bias over several quarters, and check whether the gap shrinks.
How the BarrierX method does it
The BarrierX method builds this into five gates, from Discovery Call to Contract Signed, each with the proof needed before a deal moves forward, and a Monday review that compares the CRM stage with the earned stage. The mutual action plan is where the dated next steps live.