# Pipedrive deal probability vs stage probability: why your weighted forecast is wrong

https://salesbond.app/blog/pipedrive-deal-probability-vs-stage-probability/
Published: 2026-10-03 · Updated: 2026-10-20

In Pipedrive, a deal’s own probability always beats the probability of its stage, and every stage is set to 100% by default until someone changes it. Put those two rules together and the weighted total at the top of your pipeline is wrong in a predictable direction: too high. Until you have set stage probabilities and found the deals where someone typed a number by hand, the forecast view is adding up hope, not expectation ( [Pipedrive: probability in Pipedrive](https://support.pipedrive.com/en/article/probability-in-pipedrive)).

## The forecast that looked too good

Picture a head of sales showing us a weighted pipeline that is almost the same number as the raw pipeline. Not a little lower. Almost the same. The scene is a composite, stitched together from several pipelines we have looked at, not one team’s story.

That should be impossible. If half of your open deals sit in the first two stages the weighted number ought to be a fraction of the raw one, because most of those early deals will not close. A weighted total that tracks the raw total is telling you something, and it is not that the team is about to have a record quarter.

Open the pipeline settings and the cause shows up fast. Every stage at 100% except one, where somebody typed 50% a long time ago and moved on. Then the deals themselves: a dozen with a probability set by hand, mostly 80% and 90%, mostly by one rep. In one account, several of those deals had since been moved back a stage, and their number had not moved with them.

That is the whole article in two paragraphs. The rest is how to find the same thing in about twenty minutes, and what rule to put in place so it stops coming back.

## Two probabilities, one winner

Pipedrive has two probability fields, and for any single deal only one of them is used.

Stage probability belongs to the pipeline. You set it from the pipeline view by clicking the pencil icon next to the pipeline dropdown, and you need visibility of that pipeline to do it. Deal probability belongs to one deal: you choose it in the summary section of the deal’s detail view. When both exist, Pipedrive’s documentation is blunt about which one counts: “The deal probability is always prioritized over the stage probability.” ( [Pipedrive: probability in Pipedrive](https://support.pipedrive.com/en/article/probability-in-pipedrive))

The arithmetic is simple. Weighted value is total value multiplied by probability divided by 100. If the deal has its own probability, that is the number used. If it has none, the stage probability is used. If neither is set, the weighted value simply equals the deal value. The pipeline view shows the total weighted value at the top, and the same weighted values feed the forecast view in the Deals tab ( [same Pipedrive article](https://support.pipedrive.com/en/article/probability-in-pipedrive)).

![Decision flow: deal probability if set, otherwise stage probability, otherwise full deal value; each result adds to the stage total](https://salesbond.app/wp-content/uploads/2026/10/pipedrive-deal-probability-vs-stage-probability-2.png)

So where does the deal number end up? Pipedrive puts it this way: “A deal-specific probability will override a stage’s probability, but count toward that stage’s weighted value.” For the deal itself, the manual number is not an adjustment on top of the stage number. It replaces it. For the stage, nothing is thrown away: that deal’s weighted value is added to the weighted values of every other deal sitting there, which is what the help means when it says the deal’s probability “is combined with the stage probability” ( [Pipedrive: probability in Pipedrive](https://support.pipedrive.com/en/article/probability-in-pipedrive)). One hand-typed 90% on a large deal therefore lifts the whole stage total, even if every other deal in it uses your carefully tuned stage number.

What the help doesn’t say is anything about the deal field resetting when a deal changes stage. It describes deal probability as set per deal, full stop. Treat a manual number as staying where the rep left it until a person clears it.

For the general method behind weighting a pipeline, we wrote a separate piece on [weighted pipeline forecasting](https://salesbond.app/blog/weighted-pipeline-forecast/). This article stays on the two Pipedrive fields and what they do to the number.

## The 100% default nobody changed

Pipedrive does not guess your conversion rates, so a new stage gets 100% and keeps it until a human decides otherwise ( [Pipedrive: probability in Pipedrive](https://support.pipedrive.com/en/article/probability-in-pipedrive)). That is a defensible default for software. It is a terrible default for a forecast, because 100% means every deal without its own probability contributes its full value to the weighted total. A first call that went well counts exactly as much as a signed quote.

Most teams we talk to set up their pipeline in the first week, named the stages, maybe added a couple of required fields, and never opened the stage probability setting at all. The forecast view was something they discovered months later, and by then the number it showed looked plausible enough that nobody questioned where it came from. Plausible is the dangerous part. A weighted total that is obviously wrong gets fixed. One that is merely optimistic gets presented to the owner.

So the first check is the shortest one. Open the pipeline view, click the pencil next to the pipeline name, and read the probability on each stage. If you see 100% on more than the final stage, your weighted pipeline has been a raw pipeline with a different label. If you run more than one pipeline, repeat this for each, since the setting lives on the pipeline and not on the account. Write the numbers down before you change anything; you will want them in the next two checks, and you will want to be able to explain to the team why the forecast dropped overnight after you fixed them.

## Sorting open deals by manual overrides

The second check takes a little longer and tends to be more uncomfortable, because it has names attached.

Start in the same pencil menu. Deal probability is a toggle on the pipeline (“enable deal probability”), only deals admin users can switch it on, and once it is on a deal takes a whole number from 0 to 100 ( [Pipedrive: how can I set deal probability?](https://support.pipedrive.com/en/article/how-can-i-set-deal-probability)). If the toggle is off, reps have no deal-level number to set from the deal view, and this check gets much shorter.

If it is on, open the deals list view, click the gear icon and add the “Probability” column next to stage and value ( [same Pipedrive article](https://support.pipedrive.com/en/article/how-can-i-set-deal-probability)). The list has no column for stage probability, so take that from the numbers you wrote down in the first check. Then compare each deal’s probability to the probability of its stage. Wherever the two differ, someone set the deal field, by hand or through an import (Pipedrive lets you map a spreadsheet column to the “Deal – Probability” field), and that deal has been ignoring your stage settings ever since. One blind spot: a manual number that happens to equal the stage number looks identical in this list, so this method won’t surface it.

Count those deals and add up their value. In our experience the count is small and the value is not, because the deals that get a hand-typed 90% are the big ones a rep feels good about. Those are also the deals that move the forecast the most when they slip.

Then open each [deal’s history](https://salesbond.app/blog/pipedrive-deal-changelog-close-date-audit/) to see who set the number and when. A manual probability is almost always a snapshot of of one moment: the rep came out of a strong call, opened the deal, and put 80% in the summary. Nothing in Pipedrive prompts them to revisit it. The deal then lives on for weeks, loses a champion, gets its close date pushed, maybe gets moved back a stage, and the 80% sits there untouched. If your team already pushes close dates to keep deals out of the current period, you will find the same deals showing up in both lists; we covered that habit in [what dragging a deal’s date does to the Pipedrive forecast view](https://salesbond.app/blog/pipedrive-forecast-view-dragging-deal-date/).

One more pattern worth catching here: deals where a manual probability is lower than the stage. These are rarer, and they are usually honest. A rep who sets 20% on a deal in Negotiation is telling you something they have not said out loud in a pipeline review. Ask about those before you ask about the 90%s.

## Calibrating stages against your own closed deals

Fixing the 100% default is not the same as setting the right number. The right number for each stage is the share of deals that reached that stage and eventually won, measured on your own closed deals, not on a benchmark from someone else’s industry.

1. Pick a period long enough to contain a real number of closed deals, won and lost. For a small team that is usually several months, not one.
2. For each stage, count the deals from that period that passed through it, and the subset of those that were won. Won divided by passed through is the stage’s historical win rate. How to count which deals reached each stage from a Pipedrive export, including deals that were moved backwards, is in [pipeline stage conversion rates in Pipedrive](https://salesbond.app/blog/pipeline-stage-conversion-rates-pipedrive/).
3. Put that rate next to the stage probability currently set. Where the gap is large, the stage probability is the one to change, and the direction of the gap tells you whether the weighted total has been running high or low.
4. Set the stage probabilities to the measured rates, rounded. Do this on the pipeline, not on individual deals.
5. Compare the new weighted total to what the team actually closed in the equivalent period. If it is still far off, the problem is somewhere other than probability: deal values, close dates, or deals that should have been marked lost. We go through those other sources of drift in [sales forecast accuracy for small teams](https://salesbond.app/blog/sales-forecast-accuracy-small-teams/).

Expect the weighted number to fall. Expect somebody to say the old number was more motivating. Hold the line: a weighted total built on 100% stages is a number nobody should plan against. Don’t confuse it with commit, though. If your team runs separate commit and best-case numbers, those categories answer a different question (which deals meet your rules for counting on them this period), and the weighted total sits next to them rather than replacing them; we wrote about [the difference between commit and best case](https://salesbond.app/blog/commit-vs-best-case-forecast/) elsewhere.

Recalibrate on a schedule, not when the forecast misses. Win rates drift as the team, the pricing and the market change, and a stage probability measured once and never revisited becomes the new unexamined default.

Changing stage probabilities means [editing a live pipeline](https://salesbond.app/blog/pipedrive-edit-pipeline-stages-live/), just like renaming or deleting a stage, so read what each stage edit does to your deals and reports before you do it.

## A rule for touching deal probability

Our position: deal probability should be empty on almost every deal, so the stage probability applies. A manual number is allowed only when the rep knows something the stage cannot know, and only with a note on the deal that says what that is.

The note is the rule, not the number. What we ask reps to write:

- the reason the stage probability is wrong for this deal (a verbal yes from the buyer, a budget that was cut, a competitor already chosen);
- the date, so a reviewer can see how old the override is;
- when it should be looked at again, or what event clears it.

A manual probability without a note gets reset to empty in the next pipeline review, and the rep hears why. That is housekeeping, not a reprimand. The weighted forecast is a shared number, and one unexplained 90% on a large deal changes it for everyone.

Managers get a job from this rule too: during pipeline review, look only at the deals whose probability differs from their stage. That is a short list and every deal on it has a note explaining itself or does not. It is a faster review than reading the whole pipeline, and it catches exactly the deals that will embarrass the forecast.

A disclosure, since this is the part a tool can help with: we build SalesBond, a sales management layer on top of the CRM a team already uses; we started with Pipedrive. Deal Health checks every deal in the connected CRM against more than 20 rules, including Forecast rules, assigns a health score, shows the specific violations and explains what to do. [Check your CRM with SalesBond](https://salesbond.app/). The three checks above do not need it, though. They need twenty minutes and the pencil icon.

So start with the pencil today. If more than your last stage says 100%, you already know why the forecast looked too good, and the deals list with a “Probability” column will show you where someone has been typing over it.