How to set deal probability honestly
Base probability on the stage's historical close rate, not a rep's optimism about a specific deal. A pipeline where every deal reads 80% has stopped telling anyone anything useful.
Want to see it first? Book a demo
The short answer
- Deal probability works best when it belongs to the pipeline stage and shows that stage's past close rate. Not a number a rep sets based on personal optimism about one deal.
- A default probability per stage, used the same way every time, gives a truer forecast. Numbers that shift deal by deal with each rep's mood that week don't.
- Recalculate stage-level probabilities periodically, commonly quarterly, against actual close rates. A number set once and never revisited drifts from reality as the sales process or market changes.
- Watch for most deals bunched at a high probability whatever their stage, or forecasts that keep missing. At that point the number is hope, not an estimate.
Why rep-set probability tends to drift toward optimism
When probability is set by whoever owns a deal, it tends to drift upward over time. Not because reps are dishonest. A rep working a deal is naturally more hopeful about it than the past base rate would back up. This is a well-documented pattern in forecasting. The person closest to an outcome tends to think it's more likely than it turns out to be across many similar cases.
Base probability on the deal's stage instead, with that stage's real past close rate as the default. That takes one person's bias out of the number. A deal in an early stage gets the early-stage close rate, however confident the rep feels. A deal in a late stage gets the late-stage rate. The forecast then reflects what has happened across many deals in that stage, rather than one person's read on one deal.
Deviating from the stage default should be possible but treated as the exception. It needs a specific, stated reason: a signed verbal commitment, a specific blocking objection. Not a routine adjustment. If overrides become routine, you're back to reps setting the number, with extra steps.
A stage-based probability example
| Stage | Historical close rate (example) | Default probability |
|---|---|---|
| Qualified | 10-15% of qualified deals eventually close | 10% |
| Meeting / Demo held | 20-30% | 25% |
| Proposal sent | 35-45% | 40% |
| Negotiation | 60-70% | 65% |
| Verbal commitment | 80-90% | 85% |
Illustrative figures; actual close rates should come from a specific team's own historical data, not this example.
What to actually do to set this up honestly
Pull closed-won and closed-lost deals from the last several months. Work out the actual percentage that closed from each stage. Use those figures, not round numbers that feel intuitive, as the stage defaults. Revisit the calculation on a schedule. Close rates shift as the sales process, market conditions, or the team itself changes.
Maybe you don't have enough history for solid stage close rates, say with a new sales motion or a small team with few closed deals. Start with fair guesses and mark them as temporary. Do not wait for enough data to be perfectly confident. A clearly labeled estimate that you update as real data builds up beats a random guess dressed up as fact. It is also more useful than no default at all while the team waits months for certainty.
Disclosure: SalesCrew is our product. Deal stages carry a default probability used for weighted forecasting, and stage-change history is timestamped. That gives a team the raw data to calculate its own close rates by stage rather than guessing. A quarterly review of that history keeps the probabilities tied to real results.
An inflated forecast eventually costs credibility, not only accuracy
Questions
- Should every deal in the same stage have the same probability?
- As a default, yes. Probability should show the stage's past close rate, not one rep's gut feel about one deal. Deviating from the stage default should be the exception, with a clear reason, not the norm.
- How often should stage-level probabilities be recalculated?
- Periodically, commonly quarterly, based on actual close rates from deals that passed through each stage. Probabilities set once and never checked drift from reality as your sales process, market or team changes.
- What is the sign that probability numbers have stopped being useful?
- Most deals clustering at a high probability (70-90%) whatever their stage. Or a forecast built from those probabilities that keeps missing actual results by a wide margin. Either way, the numbers have become optimism rather than a working estimate.