The short answer
Build an accurate pipeline forecast by defining each stage with buyer evidence rather than seller activity — a stage advances only when the buyer has done something verifiable, such as sharing a budget range, involving a decision maker, or accepting a proposal timeline. Then apply historical conversion rates by stage instead of rep confidence, remove any opportunity with no activity in 30 days, and review the forecast weekly against a single close-date rule.
Why forecasts miss
Almost always for one of three reasons: stages defined by what the seller did, close dates set by wishful thinking, and dead deals left in the pipeline because removing them feels like losing.
None of these is a modelling problem. They are definitional problems, and they are fixable in an afternoon.
Define stages by buyer evidence
A workable set of criteria for a services or considered-purchase business:
- Qualified — the buyer confirmed the problem, a rough budget range, and a timeframe.
- Solution agreed — the buyer has seen a scoped approach and has not objected to it.
- Decision maker engaged — the person who signs has attended a conversation.
- Proposal accepted in principle — terms and timeline discussed, contract pending.
Use history, not confidence
Replace rep-assigned probability with the actual conversion rate of each stage over the last two or three quarters. It will feel pessimistic at first. It will also be right more often.
Keep a separate commit view for deals the owner will personally stand behind — useful for management conversations, not for the model.
Weekly hygiene
Thirty minutes a week keeps a forecast usable: close or downgrade anything untouched for 30 days, push close dates only with a stated buyer reason, and check that every stage change has evidence attached. Forecast accuracy is a hygiene outcome far more than an analytics one.
Frequently asked
How accurate should a forecast be?
Within roughly 10-15% of actual on a quarterly view is a strong result for a small sales team. Consistency matters more than precision — a predictably low forecast is still useful.
Should marketing-sourced pipeline be forecast the same way?
Same stage criteria, different conversion rates. Blending sources into one rate hides which motion is actually producing.
What about long sales cycles?
Use stage-age limits rather than close dates as your primary hygiene rule. Long cycles are fine; stalled ones dressed as long ones are not.
Free GTM Diagnosis
Want this run on your pipeline?
Send two details. We come back with the two or three gaps costing you the most revenue.
No spam. No sales sequence. One reply from a human.
Prefer to talk? Book a 30-minute call