On this page
“Commit” can mean a probability, an accountability promise, or an inclusion bucket. “Best Case” can be a range, a cumulative roll-up, or an optimistic manager view. The label looks precise because it is short. The underlying object may not be.
A forecast category is a declared state on a forecasted work object. Its meaning comes from the evidence rule, inclusion rule, cutoff, owner, roll-up, and later outcome attached to it. A category name alone is not a probability.
The pipeline-hygiene article owns the data quality of opportunity records. This page owns the category object that sits on top of those records and the rule needed to compare its call with what happened later.
What do forecast categories mean?
Public explanations show why the words cannot be treated as one standard. Outreach describes Commit as an accountability signal rather than a probability estimate, while Microsoft describes Committed through a customer commitment evidence state. Two vendor explanations of one label, describing two different objects: an accountability signal and an evidence state (Outreach, 2026; Microsoft Learn, 2026).
That disagreement is not a nuisance around the metric. It is the metric boundary. A team can choose a probability interpretation, an evidence-state interpretation, or an inclusion interpretation. It must not mix them in one denominator.
| Object | Question | Do not infer |
|---|---|---|
| Forecast category | What state does this opportunity satisfy at the cutoff? | A universal close probability |
| Forecast probability | What probability is assigned under a named calibration rule? | That the category name supplies calibration |
| Forecast inclusion | Which states enter a roll-up? | That inclusion equals likelihood |
| Closed outcome | What commercial result was recorded? | That a closed result is a forecast category |
Table 1What do forecast categories mean?
Source: Table from this essay. Sources and interpretation are given in the article.
Which fields make a category reproducible?
Record the category as a state transition, not a free-text opinion:
- Object: opportunity ID, account, product, unit, and owner.
- Cutoff: the timestamp at which the category was called.
- Evidence: buyer action, contract state, next event, value, timing, or other declared signal.
- Entry rule: the minimum evidence required for the state.
- Inclusion rule: whether and how the state enters Pipeline, Best Case, or Commit totals.
- Owner: person or process responsible for the classification.
- Exit rule: promotion, demotion, omission, closure, or expiry.
- Outcome: later closed, lost, canceled, slipped, or still unobserved result.
The same record can carry a category and a separate probability. If both exist, preserve both definitions. Do not let a category label silently overwrite a model probability or a closed outcome.
What does a category worksheet look like?
The six rows below are synthetic. They contain no pipeline, seller, customer, or forecast result. They show a state rule that a reviewer could replay.
| ID | Category at cutoff | Entry evidence | Roll-up rule | Exit or review | Later outcome |
|---|---|---|---|---|---|
| F-01 | Pipeline | Problem recorded; buying path not yet evidenced | Pipeline only | Promote, hold, or omit at next review | Unobserved |
| F-02 | Best Case | Buyer next step and date recorded | Best Case under declared inclusion mask | Recheck date and owner | Slipped |
| F-03 | Commit | Customer commitment evidence and close plan | Commit under declared scope | Escalate missing evidence | Lost |
| F-04 | Closed | Contract or booked outcome recorded | Closed result, not a probability | Reconcile to source system | Won |
| F-05 | Omitted | Stale, duplicate, or outside period | Excluded with reason code | Reopen only after rule passes | Canceled |
| F-06 | Commit | Verbal signal with no dated next event | Held, not counted until rule passes | Demote or evidence the state | Open at horizon |
Figure 1The synthetic forecast-category state table
The rows are illustrative. A state name becomes useful only when evidence, inclusion, ownership, timing, and outcome remain visible.
Source: Author's synthetic state table grounded in Outreach (2026) and Microsoft Learn (2026); labels and outcomes are illustrative.
F-06 shows the control. A verbal signal may be useful context, but without a dated next event the team cannot tell whether the declared Commit rule was met. The correct disposition is a hold or demotion, not a hidden probability assumption.
How should category accuracy be measured?
First define the object being tested. If the category is a state, report later outcomes by category and cutoff cohort. If it is intended to be a probability, test calibration against the later outcome with a declared horizon and denominator. If it is an inclusion flag, reconcile the included amount to the roll-up rule.
Outreach describes one accuracy method using the difference between the Day One Commit call and cumulative Closed results at period end, and it identifies Closed as the denominator in that calculation (Outreach, 2026). That is a definition of one method, not evidence that all systems use the same object.
Avoid the common substitution:
Commit amount / pipeline amount
is a coverage or composition ratio unless the team has declared something else. It is not a calibration test, a win rate, or proof of forecast skill.
What are forecast categories not?
They are not a universal probability scale, a substitute for opportunity evidence, or a guarantee that a deal will close. They are not the same as sales stages, although a stage may supply evidence for a category. They are not a result because the result occurs after the cutoff.
A forecast category earns its name through a stable decision rule. The label can travel across teams only after the rule, denominator, roll-up, and outcome definition travel with it.
The forecast-override page extends this state logic with the documented change made after the original forecast.
References
- Microsoft Learn. (2026). Capture forecast category for opportunity. Dynamics 365 Sales documentation. Source page
- Outreach. (2026). Sales forecast categories explained for RevOps teams. Outreach resources. Source page
