Stage probability calibration sheet
The 10, 20, 40, 60, 80 percentages in most CRMs were typed in during implementation and never checked. Weighted pipeline built on them is a guess dressed up as a calculation. This sheet replaces each default with the stage-to-close rate your own resolved deals actually produced, by count and by value, and tells you when the sample is too small to trust.
Download it
What you get
- A cohort method for measuring stage-to-close conversion without counting open deals as losses
- A calibration table comparing default CRM percentages with observed rates by count and by value
- A weighted pipeline comparison showing how much the defaults overstate or understate the number
- A minimum sample-size caution with a rule for when to merge segments or widen the window
- A recalibration checklist for the quarterly refresh
Who it's for
- RevOps analysts who own weighted pipeline and forecast models
- CRM admins asked to change stage probability fields
- Sales leaders who suspect weighted pipeline is too optimistic
What's inside
5 sections, each listed with its own size. The first is previewed below; the remaining 4 come with the download, with blank rows ready to fill in.
| Section | Size | Previewed |
|---|---|---|
| Calibration table | 11 columns, 5 worked example rows | Below |
| Formulas | 3 columns, 5 worked example rows | In the download |
| Weighted pipeline: default vs calibrated | 6 columns, 6 worked example rows | In the download |
| Sample size and what the example shows | Guidance notes | In the download |
| Quarterly recalibration checklist | 8-point checklist | In the download |
Preview of section 1
Calibration table
Worked example: opportunities that entered each stage in a closed-out four-quarter window. Replace with your own stage history export.
| Stage | Default CRM % | Entered stage | Won | Lost | Still open (excluded) | Observed rate by count | Resolved value | Won value | Observed rate by value | Calibrated % (rounded) |
|---|---|---|---|---|---|---|---|---|---|---|
| Discovery | 10% | 400 | 60 | 320 | 20 | 15.8% | 12,000,000 | 1,500,000 | 12.5% | 13% |
| Qualified | 20% | 250 | 55 | 185 | 10 | 22.9% | 8,000,000 | 1,400,000 | 17.5% | 18% |
How to use it
1. Pick a cohort window that has finished resolving
Use opportunities that entered each stage in a window old enough that nearly all of them are now won or lost, typically four to seven quarters back for a mid-length cycle. Using last quarter's entries undercounts wins that have not closed yet.
2. Count every deal that ever entered the stage
Use stage history, not current stage. A deal that went from Proposal to Closed won counts as a win for Discovery, Qualified, Solution, and Proposal. Deals that skipped a stage do not count for that stage.
3. Exclude still-open deals rather than counting them as losses
Observed rate = Won / (Won + Lost). If more than a small share of the cohort is still open, the window is too recent. Move it back a quarter.
4. Use the by-value rate for weighted pipeline
Weighted pipeline is a dollar figure, so it needs a dollar-based rate. If the by-value rate is well below the by-count rate, large deals win less often than small ones, which is worth raising with sales leadership on its own.
5. Check sample size before you publish
If a stage in a segment has too few resolved deals, merge it with a neighbouring segment or widen the window. Do not publish a probability that one deal can move by several points.
Frequently asked questions
What is opportunity stage probability?
The likelihood that an opportunity in a given stage eventually closes won. CRMs use it to compute weighted pipeline (amount x probability). It should be measured from your own history, not left at the implementation defaults.
How do you calculate stage probability from historical data?
Take all opportunities that entered the stage in a window old enough to have resolved, then divide the number (or value) that closed won by the number (or value) that closed won or lost. Exclude deals that are still open.
Should stage probability be by count or by value?
Use by value for weighted pipeline, since the output is dollars. Track by count as well; a big gap between the two means large deals convert worse than small ones.
How often should stage probabilities be recalibrated?
Quarterly is a common cadence, or after any change to stage definitions, pricing, or segment mix. Recalibrating monthly on small samples just adds noise.