Why 'Commit' Doesn't Mean the Same Thing at Any Two Companies
There's no shared definition behind commit, best case and pipeline — which is why the 'industry average forecast accuracy' stat you keep citing is fiction.
Somebody in your org has quoted the stat: average enterprise sales forecast accuracy sits somewhere between 60% and 80%, depending which analyst report you pulled it from. It gets repeated in QBRs, cited in RevOps decks, used to benchmark your team against "the industry." It is, as far as I can tell, closer to astrology than analytics. Not because the underlying data collection is bad — because the word "commit" in that dataset means something different at every single company that contributed to it, and nobody adjusted for that before averaging them together.
I've built or inherited forecast categories at four different companies. I have never seen two definitions of "commit" that matched. At one, commit meant the rep was personally confident above 80%. At another, it meant a manager had reviewed the deal and approved the category — a completely different signal, since managers vary wildly in how sceptical they are. At a third, commit required a specific, named evidence checklist. Only the third one produced a number worth reporting to a board.
The categories were never standardised — and nobody tells you that
Pipeline, best case, commit, closed-won: these terms read like a taxonomy. They're not. There's no governing body, no ISO standard, no shared glossary that CRM vendors, sales methodologies, or companies agree on. Salesforce ships default stage names. MEDDICC gives you qualification criteria, not forecast categories. Command of the Message tells you how to run discovery, not when a deal graduates to commit. Every company is left to invent its own line, usually by committee, usually without writing down what evidence actually has to exist for a deal to cross it.
Here's what that produces in practice — three patterns I've seen repeatedly, lightly composited:
| Company type | What "commit" actually meant | What it should have meant |
|---|---|---|
| Series B SaaS, ~40 reps | Rep says "I'm confident" on the forecast call | Nothing verifiable — pure sentiment |
| Mid-market services firm | Deal has a signed proposal on file | Proposal sent is not the same as proposal reviewed by the economic buyer |
| Enterprise software, 300+ reps | Manager has personally reviewed and approved | Consistency depends entirely on how sceptical that particular manager is |
None of these are wrong, exactly. They're just incompatible with each other, and none of them is what a board or a public-market analyst assumes "commit" means when they hear the word — which is closer to "this money is coming in, full stop."
Why this makes benchmarking fiction
If Company A's commit closes at 55% and Company B's commit closes at 92%, and someone averages the two "commit accuracy" figures into an industry number, that figure describes nothing real. It isn't measuring forecasting skill. It's measuring how loosely two different companies used the same word. Layer in that most public "industry average" forecast-accuracy stats are self-reported, aggregated across an unknown number of companies with unknown category definitions, and you get a number that's less useful than no number at all — because a bad number gives false confidence, and no number at least prompts someone to ask a question.
The tell is simple: ask whoever quotes the industry stat to define, precisely, what evidence a deal needs to sit in commit. If the answer is a feeling rather than a checklist, the benchmark is decoration.
What an actual definition looks like
The fix isn't complicated, it's just rarely done properly. A forecast category needs a binary evidence threshold, not a confidence level. Here's a version I've implemented that held up under audit:
Commit — all four must be true:
- Economic buyer identified by name and personally engaged (a call, email, or meeting — not "the champion says they're on board")
- A second stakeholder is engaged independently of the champion
- A mutual close plan exists with dates both sides have acknowledged
- Procurement, legal, or security review has started, not just been mentioned as upcoming
Best case — two or three of the above are true, with a dated next step on the calendar.
Pipeline — qualified opportunity, fewer than two of the above true.
Notice none of these criteria ask how the rep feels. That's deliberate. Confidence is the thing forecast categories exist to remove from the equation, not encode into it.
What to do with this on Monday
- Pull your current category definitions — the ones actually written down, if any exist — and check whether a single one references a piece of evidence rather than a feeling.
- Rewrite each category as a checklist, not a description. If you need a starting point rather than building the thresholds from nothing, the Forecast Category Definitions Cheat Sheet lays out working versions across a few qualification frameworks.
- Re-run last quarter's forecast against the new definitions before you present the new categories as a change. You'll usually find a chunk of "commit" was never commit at all, and it's better you find that before your board does.
- If you want a live view of what's genuinely evidenced versus what's a rep's gut feel, run current pipeline through something with the criteria built in rather than a colour-coded spreadsheet — the Deal Review Checklist is built for exactly that pass.
None of this makes your forecast accuracy number go up next quarter. It'll probably go down, because you'll be honest about what was never real commit to begin with. That's the point. A forecast that's finally measuring something is worth more than one that's comforting and empty.