How to Set Deal Stage Probabilities From Your Own Data

To set deal stage probabilities, ask one question per stage: of the deals that reached this stage, how many eventually closed as won? Use your own history. With none, set conservative guesses, label them as guesses, and replace them after a quarter of real outcomes.

How to set deal stage probabilities without lying to yourself

Ten per cent. Twenty-five. Fifty. Seventy-five. Ninety.

You've seen that ladder. It comes pre-installed in most sales tools. It looks scientific. It describes nobody's business.

Using it is the single most common way a solo operator ends up with a confident forecast and a thin bank balance. The percentages aren't yours. They didn't come from your buyers, your price point, your sales cycle or your habit of telling yourself a deal is fine.

So throw them out. Here's how to set deal stage probabilities that replace them, and it takes less time than arguing with the defaults.

Why probability belongs to the stage, not the deal

Ask yourself how likely a particular deal is and you'll get a mood. A good call yesterday means high. A silent inbox means low. You're reading your last conversation, not the odds.

And you're the worst person to ask. The one closest to a deal is the one most invested in it, and when you sell alone, that's you every single time.

The stage doesn't care how you feel. Of all the deals that ever reached it, some number closed. That's a fact about your business, and it's far more stable than any hunch. It'll be wrong about one deal. It'll be surprisingly right about twenty.

That's the whole trick. Take the mood out of the arithmetic by attaching the probability to the stage, and let the history of the stage supply the number.

Free guide · 20 pagesForecast the number you steer by.The Sales Forecast Playbook: stop adding up deals that feel likely.Get the free guide →

First, fix the stages

You can't calibrate a stage that means different things on different days.

Each stage needs an exit criterion that's a fact: something the buyer did, which you can check. Discovery call held. Decision maker engaged. Proposal reviewed with the buyer. Verbal agreement. Paperwork in procurement.

Deal stage probabilities only work on stages that are facts. Compare two ways of naming the same stage:

If a stage can be entered because you're hopeful, its probability will drift upwards until it means nothing. This is why the stage list comes before the percentages. Get it wrong and every number after it is built on sand. The wider build sequence is in how to build a sales forecast.

The calculation, if you have history

One question per stage. No spreadsheet gymnastics.

1. List every deal that ever reached the stage. Won, lost, still open if they're old enough to judge.
2. Count how many of them eventually closed as won.
3. That share is the stage's probability.

Say you sell a productised service and look back at everything that ever reached "proposal reviewed with the buyer". If roughly one in three of those closed, your probability for that stage is about one in three. Not what the software says. Not what feels right after a good meeting. What happened.

Expect an uncomfortable discovery. Early stages will be lower than the defaults suggested, and you'll find a stage or two where deals pile up and almost never convert. That's not a failure of the method. It's the method working. You've just found where your pipeline leaks.

The workaround, if you have none

Most solo operators reading this don't have clean history. Fine.

Set conservative guesses. Write down that they're guesses. Put a date next to them for when you'll replace them. Then get on with it.

Three rules for guessing:

• Err low. A forecast that's pleasantly surprised you is cheaper than one that burned your cash.
• Label every guess. Future you will want to know which numbers were measured and which were invented on a Tuesday.
• Replace them as soon as a quarter of real outcomes exists.

A probability you calibrated badly is still better than one you borrowed. Yours can be corrected. Borrowed ones just sit there, looking authoritative.

A worked example, in words

Say you sell a productised audit to small firms. You look back over the last year of deals.

You find that most deals which reached the discovery call went nowhere. Many that reached the proposal stage went quiet. But almost everything that reached verbal agreement signed, with a few stalling at paperwork.

Now compare that with the software's ladder. It would have given proposal a number far more flattering than your own record supports, and given verbal agreement less credit than it deserves. So your forecast was too high in the middle and too low at the end. Too high is the dangerous direction. It's the one that talks you into a contractor.

Notice what you did there. You didn't build a model. You counted, once, and the pattern showed itself. The count is the work. Everything after it is multiplication.

Free assessmentHow trapped are you, really?Take the Corporate Suffocation Index, the free solopreneur readiness assessment.Take the assessment →

Two stages that need no guess

Won and lost don't take a probability. They're outcomes. But they need to exist, clearly defined, or the whole calculation falls over.

A deal is lost when someone says no, or when it has gone silent past a limit you set in advance. Decide that limit now. If you wait until a deal is making you uncomfortable, you'll always find a reason to give it another month.

This matters for the maths. A pipeline that never marks anything lost has stage numbers that look healthy and mean nothing, because the deals that failed never counted against the stage they died in. You can't measure how many reached a stage and closed if the ones that didn't are still sitting there pretending to be open.

Where the numbers go wrong

Small samples lie. With a dozen open deals, a stage probability is a polite fiction, and one large deal can outweigh everything else. Two defences.

First, name big deals individually. Don't let one hide inside a weighted total. Treat it as its own line with its own judgement.

Second, remember that age changes the maths. Every business has a typical sales cycle. A deal that has been open for twice that length is no longer an average deal at its stage. It's stuck, and the stage probability no longer describes it. Old deals inflate a weighted pipeline more quietly than anything else. That's one reason a weekly session earns its place, and how to run a weekly pipeline review shows how to catch them.

What the probabilities are, and are not, for

Once each stage has a number, multiply every open deal by it and add them up. That's your weighted pipeline. It tells you what the pipeline is worth on average, over time.

It does not tell you what this month will bring. A month contains too few deals for averages to behave. That's a different number, and it needs a different method, set out in weighted vs committed pipeline.

Probabilities are for the long range. Don't ask them to do the near-term job.

Free guide · 20 pagesForecast the number you steer by.The Sales Forecast Playbook: stop adding up deals that feel likely.Get the free guide →

Recalibrate your deal stage probabilities every quarter, in writing

The numbers you set today are the first draft. At the end of each quarter, compare what each stage was supposed to convert with what actually did. If the stage said one in three and delivered one in five, change it. Even when reality is unflattering. Especially then.

Three signs your probabilities have stopped being honest: they haven't changed in a year, every deal seems to close in the last week of the period, and close dates move while stages never do. Any one of those means the number has drifted from the deals underneath it. The fix is the calibration, not a prettier spreadsheet.

Do this properly and your deal stage probabilities become the most reliable thing in your business. They're boring. They're yours. And they don't care how your week went.

Free assessmentHow trapped are you, really?Take the Corporate Suffocation Index, the free solopreneur readiness assessment.Take the assessment →

Frequently asked questions

Should I use the default probabilities in my sales software?

No. Defaults describe nobody's business. They feel authoritative and drift further from reality every quarter they go unchecked. Set your own from the share of deals that reached each stage and eventually closed.

What if I have no sales history at all?

Set conservative guesses, label them as guesses, and put a date on replacing them. Once a quarter of real outcomes exists, swap in measured numbers. A badly calibrated probability of your own beats a borrowed one.

How often should I change my deal stage probabilities?

Review them at the end of every quarter. Compare what each stage was meant to convert with what actually converted, and adjust to match. Changing them mid-quarter because of a good or bad week just puts your mood back in.

Can I give one deal a different probability from its stage?

Avoid it. The point is to take your judgement out of the arithmetic. If a deal is genuinely special, name it individually and treat it as its own line, rather than quietly overriding the stage.

The full method, with the checklists, is in The Sales Forecast Playbook, a free 20-page playbook.