Deal Value, Probability and a Forecast You Can Trust

Ask most small business owners what will close this month and you get a number that is part memory, part hope and entirely unfalsifiable. A pipeline can do better than that, but only if the two fields underneath the forecast are honest. Our guide to pipelines, stages and the deals you are actually going to close treats value and probability as the pair that makes the whole board either useful or decorative.

Value: what you will invoice, not what you hope for

The most common error is entering the best-case number. Someone asks about a kitchen; you enter the price of the kitchen you would like to sell them rather than the one they described. Multiply that across thirty deals and your pipeline is showing a business twice the size of the real one.

Enter the value of the work actually discussed. If it is genuinely a range, enter the low end — a pipeline that under-promises produces good surprises, and a pipeline that over-promises produces staffing decisions you regret. Update the figure when the quote goes out, because that is the first moment you have a real number.

One value, defined once

Decide whether deal value means the first invoice or the whole relationship, and apply it everywhere. For a one-off job they are the same. For a subscription or a retainer they are wildly different, and a board mixing both cannot be summed.

The usual answer is to record the first year's value on the deal and track the longer horizon separately, for the reasons set out in understanding and growing customer lifetime value. What matters is that everyone uses the same convention.

“

Probability is not a feeling about the customer. It is the percentage of deals that historically closed from this stage.

— be digital ai team

Derive probability from your own history

Most teams assign probability by instinct, and instinct is systematically optimistic — nobody marks their own deal at twenty per cent. The alternative takes an hour and lasts a year: count what actually happened.

1. Take every deal from the last twelve months, won and lost.
2. For each stage, count how many deals passed through it and how many of those eventually closed.
3. That ratio is the probability for that stage. If forty of your last hundred quoted deals were won, quoted is forty per cent.
4. Apply it to every deal in that stage, regardless of how the last conversation felt.

This will be uncomfortable the first time. Most businesses discover their early stages convert far worse than assumed, which is not bad news — it is the first honest picture of the funnel they have had.

Resist per-deal adjustments

The temptation is to say this particular customer is different, and nudge one deal to eighty per cent. Occasionally that is right. Usually it is the optimism the stage-based number exists to remove, and once one deal is hand-adjusted the whole forecast is a matter of opinion again.

If a deal genuinely deserves a different probability, that is usually a sign it has reached a stage you have not defined — a verbal yes with paperwork pending, for example — which is a stage design question rather than a percentage question, covered in designing pipeline stages that match how you sell.

Add the date, or the forecast is meaningless

Value times probability tells you how much is in play. It does not tell you when, and "when" is the entire reason anyone asks. Every deal needs an expected close date, set by the customer's timeline rather than the end of your quarter.

When one does turn out to be dead, record why — the practice in learning from lost deals is what stops the same forecast error repeating. Deals whose expected close date has passed are the most useful list on the board: either the date was wrong, or the deal is dead and nobody has said so. Reviewing that list weekly, as part of the routine in running a weekly business review with your copilot, keeps a forecast connected to reality.

Check it against what happened

Once a month, compare last month's forecast with what actually closed. If you consistently forecast more than you deliver, your probabilities are too generous or your values are aspirational, and you can fix the specific one that is wrong. A forecast nobody checks afterwards is a wish; a forecast that gets compared to reality every month becomes accurate within a quarter, almost regardless of how bad the first one was.