How to build a sales forecast your board will believe

Most small-company forecasts are optimistic, and they are optimistic in a consistent way. That consistency is useful: a predictable bias is correctable.

Where the optimism comes from

Stage probabilities are usually set once, by feel, at the point the CRM is configured — and then never revisited. Someone decides Proposal means 50%, and it stays 50% forever regardless of whether proposals actually close half the time.

Use your own history instead

You do not need a data team. Take every deal that reached a given stage in the last twelve months and count how many were eventually won. That ratio is your probability for that stage. It will almost certainly be lower than the number currently in your CRM.

1. Export closed deals for the last year, won and lost.
2. For each stage, count deals that ever reached it.
3. Count how many of those were won.
4. Won ÷ reached = the real probability for that stage.
5. Put those numbers into your pipeline settings.

Then fix the stage definitions

If a stage produces a probability wildly different from what you expected, the stage is usually badly defined rather than the deals being unusual. Stages that describe your activity — "proposal sent", "demo booked" — predict poorly, because doing a thing is not the same as the buyer committing to anything.

Record why you lose

A forecast tells you what will happen. Loss reasons tell you why, and they are the input that actually improves the number next quarter. Make the field required, keep the list short enough that people pick honestly, and read it once a month.