How do small teams forecast? Commit, best case and pipeline explained
Forecast categories sort your open deals by how sure you are they will close this period. Pipeline is early and uncertain, best case might land this period if the open questions break your way, and commit is the deals you'd bet your quarter on. A small team can forecast with three columns in a spreadsheet: list open deals, tag each one, add up each column, and compare the sums to the quota you have left. The category names are standard in CRMs, but what each one means is up to your team, so write the rules down.
Updated
What do the forecast categories mean?
Salesforce's help describes a forecast category as the category an opportunity is assigned to, based on its stage, and lists the standard ones as Pipeline, Best Case, Commit, Omitted and Closed. HubSpot's default deal properties include a forecast category with the values Not forecasted, Pipeline, Best case, Commit and Closed won. Neither page gives a rule for what belongs in each bucket, so the definitions below are ours.
| Category | Our definition | A test you can use |
|---|---|---|
| Pipeline | A real opportunity that could close this period but isn't likely yet | A buyer has agreed there's a problem; no decision date or signer yet |
| Best case | Might land this period if the open questions resolve well | A signer is identified and a decision date falls inside the period |
| Commit | You'd bet your number on it | The buyer has confirmed the date and next step, the signer is engaged and the paperwork is moving |
| Closed (won) | Signed | Counted already, not forecast |
| Omitted / not forecasted | Leave it out of the forecast | Stalled, or can't close in this period |
Write the tests down and agree them as a team. If every rep decides on their own what "commit" means, the totals mean nothing. Salesforce lets admins rename categories and add one, and HubSpot supports custom categories, so you can match your own names.
How do you build a forecast with three columns?
- List every open deal with its value, stage, expected close date and the signer, from your CRM or a spreadsheet.
- Tag each deal with one category, using your written tests. Be strict. A deal that fails the commit test is best case.
- Add up each column. You get a commit total, a best case total and a pipeline total.
- Compare to quota left. Subtract what you've already closed from quota. See how much pipeline you need.
- Review weekly. Move deals between categories, and note why they moved.
A worked example, invented for illustration. Quota left this quarter: $100,000. Open deals:
| Category | Deals | Value |
|---|---|---|
| Commit | 2 | $36,000 |
| Best case | 3 | $54,000 |
| Pipeline | 14 | $210,000 |
Commit covers 36% of what's left. Commit plus best case is $90,000, or 90%. If every commit and best case deal closes, you'd still be $10,000 short and need something from the pipeline column. That's useful on its own: it tells you to build more pipeline now, not at the end of the quarter.
How do you turn categories into a number?
Two ways. The first is to read the columns as ranges: commit is your floor, commit plus best case is your stretch. The second is to put a weight on each column. Weights must come from your own history, and the ones below are invented, not benchmarks:
| Category | Value | Illustrative weight | Weighted value |
|---|---|---|---|
| Commit | $36,000 | 90% | $32,400 |
| Best case | $54,000 | 40% | $21,600 |
| Pipeline | $210,000 | 10% | $21,000 |
| Total | $75,000 |
Our math: 36,000 × 0.9 = 32,400; 54,000 × 0.4 = 21,600; 210,000 × 0.1 = 21,000. That's $75,000 against $100,000 left, a $25,000 gap.
To find your real weights, look at deals tagged each way at the start of past quarters and count how many closed. If you have fewer than about 30 deals per category, the weights are rough, so use ranges instead. For win rate by stage, see what a good B2B win rate is.
Can AI forecast for a small team?
Limits first, because they decide everything else.
- Small sample. A team closing a few deals a quarter gives a model almost nothing to learn from. One big win or loss swings every pattern it finds.
- Few deals. With a dozen open deals, you can read every one yourself before the weekly review. A score on top of that adds a number, not information.
- Dirty stage data. A model trusts your stages and close dates. If reps skip stages or leave dead deals open, it learns your mess and hands it back looking precise.
- It can't know what the buyer hasn't said. The signer's real timeline, a budget freeze, a competitor in the room: none of it exists until someone asks and writes it down. No tool predicts a conversation you haven't had.
So don't buy a prediction engine for a five-person team. Use AI as a second pair of eyes on the three columns you already keep:
- Spot stale deals. Paste your open deals with last-activity dates and ask which ones have gone quiet. Quiet deals don't belong in commit.
- Summarize deal notes. Have it pull the buyer's confirmed date, next step, signer and paperwork status out of messy notes, so you can run the commit test without rereading a thread.
- Check the categories against evidence. Ask it to flag any commit deal whose notes miss part of the commit test (confirmed date and next step, engaged signer, paperwork moving), and any best case deal that already passes all of it.
Here are my open deals with notes: [PASTE DEALS]. My written tests: commit needs [COMMIT TEST]; best case needs [BEST CASE TEST]; pipeline needs [PIPELINE TEST]. For each deal, quote the evidence for its current category, or say none found. List deals that look mis-tagged.
You make the call on every flag. The AI only points at deals to question, and the weekly review is where you move them.
How is this different from coverage and velocity?
- Coverage asks how much pipeline you need, using your win rate. See pipeline coverage.
- Velocity asks how fast pipeline turns into revenue. See the sales velocity formula.
- Forecast categories ask which of today's deals you believe, and how many dollars each group represents.
You need all three. Coverage plans the pile, categories grade it, and velocity measures how fast it moves.
If you're short on deals, the Pipe Playbook calculator estimates sends, replies and meetings, so you can see how much outbound it takes to fill the pipeline column.
Which tool should you use?
Any CRM that lets you tag deals with a category and sum by category works. HubSpot's setting for automating forecast categories updates the category as deals move stages, per its knowledge base, and Salesforce ties each category to a stage. Our CRM comparison is on the best CRM for founder-led sales. For the weekly activity numbers that create the pipeline, see SDR metrics to track every week.
When should you skip each method?
- Forecast categories in a CRM: skip them if you have fewer than a dozen open deals. A spreadsheet with a "confidence" column gives you the same answer.
- Automatic category by stage: skip it if your stages don't match how deals really move. A deal can sit in a late stage and still be stuck.
- Weighted forecasts: skip them until you have your own closed-deal history, because invented weights give a precise-looking number you can't defend.
- Ranges (commit and commit plus best case): skip them if your board or boss wants one number. Give the range and say which end you'd bet on.
Sources
Research and vendor pages accessed October 1, 2026.
- Salesforce Help, Customize forecast categories: https://help.salesforce.com/s/articleView?id=sf.forecasts3_customizing_forecasts_categories.htm&type=5
- HubSpot Knowledge Base, HubSpot's default deal properties: https://knowledge.hubspot.com/properties/hubspots-default-deal-properties