Every CRO knows the categories: Commit, Best Case, Pipeline. Every sales team uses slightly different definitions for them. And almost every forecast meeting involves someone arguing about which category a deal belongs in based on vibes, rep credibility, and deal age. What almost never happens is a systematic, mathematically grounded approach to the commit/best-case distinction — one that CROs can actually defend to their CFO or board when the quarter closes short.
This isn't an argument for replacing human judgment in forecasting. Experienced CROs carry deal-pattern recognition that no spreadsheet will replicate. But judgment applied without a mathematical framework produces forecasts with unnecessary variance. The question is: what does the math actually look like, and how do you build a commit/best-case distinction that means something consistent across your sales team?
What "Commit" Actually Means (And Doesn't)
In most sales organizations, Commit means "the rep is putting their credibility behind this deal closing this quarter." It's a statement of rep confidence, not a mathematical assertion about probability. This is the root of the problem. When Commit is a credibility signal rather than a probability statement, it means different things from different reps. A rep with a history of underselling their pipeline uses Commit only for absolute certainties. An optimistic rep uses Commit for anything they feel good about. The CRO has to mentally adjust both before rolling up the number — which introduces exactly the kind of judgment bias that makes forecasts noisy.
A better definition of Commit — one that the math can support — is: a deal where the current behavioral and process signals indicate greater than X% probability of closing in the current period, based on observable deal evidence, not rep sentiment. The X depends on your business and your CRO's risk tolerance. At most organizations I've worked with, 70-75% is a reasonable threshold. Anything above that, the deal goes in Commit. Between 40-70%, Best Case. Below 40%, Pipeline.
The question then is: how do you calculate the probability, and what evidence do you use?
The Inputs That Actually Move Deal Probability
If you want a probability estimate you can defend, you need it grounded in factors that have demonstrable correlation with close outcomes. Based on what we observe in behavioral signal data across growing B2B sales teams, the strongest predictors of whether a deal closes in a given period are:
- Prospect-side engagement trend: Is the champion's response latency trending up or down? Are more or fewer stakeholders attending calls compared to three weeks ago? Engagement is directionally predictive — increasing engagement correlates with near-term close, decreasing engagement correlates with slip or stall.
- Confirmed economic buyer access: Has the rep had a direct conversation with the economic buyer, or are they working entirely through the champion? Deals where the economic buyer is confirmed and engaged have substantially higher close rates than deals where the AE has never spoken to the person who controls the budget.
- Mutual close plan existence and currency: Is there a shared document (or at least a shared understanding) of the remaining steps to close, with dates? A mutual action plan with prospect-acknowledged milestones is one of the clearest signals that a deal is in real close-stage motion.
- Procurement/legal status: Has the deal entered formal review stages? What's the lag between the rep's expected close and the actual contract execution timeline based on the prospect's procurement process?
- Deal age vs. median cycle: How does this deal's age compare to your historical median cycle time for similar deal sizes? Deals significantly older than median have lower close rates in any given period, regardless of stage.
Building the Commit/Best Case Calculation
You can build a simple probability model from these inputs without needing a data science team. Assign each factor a weight based on how predictive it's been in your historical data. If you don't have historical data, use the order above as a reasonable starting point — economic buyer access and engagement trend carry the most weight; deal age carries the least but is a meaningful modifier.
Then score each deal in your pipeline: confirmed economic buyer engagement is worth more probability points than a champion-only relationship. A mutual close plan with acknowledged dates is worth more than an informal verbal agreement. Increasing engagement trend adds probability; decreasing trend subtracts it. Deal age past 150% of median cycle time applies a multiplier that reduces probability across the board.
Sum the weighted scores. Deals above your Commit threshold go in Commit. Deals in the middle band go in Best Case. Deals below 40% aren't being forecast as anything meaningful in the current quarter — they're future pipeline.
Will this model be exactly right? No. Will it be more consistently right than the current system of rep-credibility-adjusted vibes? Almost certainly yes. The goal isn't a perfect probability estimate — it's a consistent, defensible one that means the same thing from rep to rep and from quarter to quarter.
The Expected Value Calculation CROs Should Be Running
Once you have probability estimates per deal, the CRO's forecast becomes straightforward in concept, if not always in execution. Expected value per deal = deal size × close probability. Sum the expected values across all deals in your pipeline. That's your signal-weighted forecast.
Compare this to your stage-weighted forecast (which applies historical close rates by stage) and your rep-sentiment forecast (the traditional roll-up). In a mature forecasting environment, you use all three as cross-checks. If your signal-weighted number is significantly lower than your stage-weighted number, it means the behavioral health of your pipeline is weaker than its stage composition suggests. If your signal-weighted number is significantly lower than your rep-sentiment roll-up, it means reps are overweighting their commit confidence relative to the behavioral evidence.
The discrepancy between these three numbers is itself informative. A CRO who can say "our signal-weighted forecast is $2.1M, our stage-weighted is $2.4M, and our rep-sentiment roll-up is $2.7M" is in a much better position than one who presents a single number and hopes it's right. The spread tells the board something real about where uncertainty lives in the pipeline.
The Sandbagging Adjustment Problem
One legitimate objection to probability-based forecasting is that reps who know their deal scores are visible will manage their behaviors to influence the scores rather than improve the deal. The rep who knows that responding to champion emails faster improves their score might nudge their champion to respond faster without doing anything that actually improves deal health.
This is a real risk, and I don't want to minimize it. But it's worth comparing it to the risk we already accept: a system where reps consciously manage which deals they put in Commit based on career protection instincts, creating structural underforecasting that the CRO has to manually adjust. At least in a signal-based system, the behaviors you're incentivizing (faster engagement, stakeholder breadth, mutual close plans) are actually correlated with better deal outcomes. The current system incentivizes sandbagging, which is correlated with nothing except CRO frustration.
What the Math Doesn't Solve
Mathematical forecasting models don't solve the problem of deals that die for reasons that aren't reflected in behavioral signals. A champion who's been highly engaged can still leave the company. A deal that looked imminent can get pulled by a CFO decision to freeze discretionary spend. A competitor can undercut you on pricing at the last minute.
What the math does is reduce the variance that comes from the sources you can measure and model. It doesn't eliminate uncertainty; it concentrates your residual uncertainty in the genuinely unpredictable domain rather than letting it bleed into the domain you should already have a handle on. A CRO with a signal-grounded forecast model who misses a quarter because of an unexpected freeze is in a different position from one who missed because the model was structurally noisy to begin with. The explanation is different. The credibility conversation with the board is different. And over time, the track record is different too.
Commit should mean something. Best Case should mean something different. The math exists to make those categories real rather than rhetorical — and to give the CRO a number they can actually defend when it's time to close the books.