Every forecast call runs the same ritual. Someone reads out the commit number. Someone else asks if it's "solid." Everyone in the room nods, because nodding is what gets you out of the meeting fastest. Then the quarter ends, the number was wrong again, and someone opens a spreadsheet to work out why.
Here's the question that never gets asked in that room: why would anyone on that call tell you the truth?
Not because your reps are dishonest. Because you've built a forecast process where an honest number and an optimistic number cost the person reporting them completely different amounts. One gets you a hard conversation today. The other gets you a hard conversation in twelve weeks, by which point it's "the market" or "procurement" or "timing" — never the person who called it commit. Guess which one people choose.
Your forecast isn't broken. It's normal
Clari Labs surveyed 400 CROs, VPs of sales and revenue operations leaders at enterprise organisations in North America in late 2025. Eighty-seven percent had missed their 2025 revenue target — despite record spending on AI tools meant to fix exactly this problem. Fifty-five percent said they were getting conflicting pipeline signals from disconnected data sources. Thirty-nine percent only recalibrate their forecast models weekly or monthly, meaning a deal can drift for weeks before anyone officially notices.
Separately, Gartner's research on B2B sales forecasting found fewer than half of sales leaders and sellers report high confidence in their own organisation's forecast accuracy.
Read those two findings together. If 87% missed target and less than half trust the process that produced the number, you're not looking at a scatter of unlucky companies with sloppy reps. You're looking at the default condition of the profession. When a failure rate is that widespread, the explanation can't be individual. Nobody trains 87% of the market's sales leaders badly on the same afternoon. Something structural is producing this result, on schedule, everywhere.
So drop the theory that your forecast is uniquely broken. It isn't. That's actually the useful part — it means the fix isn't a better rep or a stricter manager. It's somewhere in the machine.
"Commit" was never a promise
Here's the sentence that should unsettle you most: nobody has a 100% conversion rate on committed deals. Not the best teams in the business. Clari's own customer success data puts top-performing teams at roughly 80% conversion on what they call "commit." Weaker teams land closer to 60%. Even at the high end, one in five deals your best people swear is locked in doesn't close this quarter.
Sit with that. "Commit" isn't a fact. It's a working guess, dressed in a word that sounds like a fact because a working guess is an uncomfortable thing to say out loud in a leadership meeting. Your CRO doesn't want to tell the board "we think we'll probably land 80% of what we're calling committed." So the language gets firmer than the reality underneath it, at every level of the reporting chain, until the number that reaches you looks solid and was never solid at all.
This is where most forecast-accuracy advice goes wrong. It tells you to tighten your stage definitions, clean your CRM, add more mandatory fields. Useful, but beside the point. You can define "commit" with perfect precision and still get an 80% hit rate, because 80% is what committed genuinely means when honest humans are estimating the future. The problem isn't the definition. It's that everyone upstream of you has quietly agreed to speak about probability as if it were certainty, because certainty is what gets rewarded in the room.
The two-way lie your incentives are training
Sandbagging gets talked about like a rep problem — someone deliberately holding a strong deal out of the forecast to protect next quarter's number, or to guarantee they beat plan and look good against a lower bar. It's real, and it's rational. If your comp plan punishes a rep for smashing quota by quietly raising their target next period, you've taught every good rep on the team to under-report. You built that behaviour. They just responded to it.
But the more damaging version runs the other way, and it usually starts above the rep, not below them. A VP under board pressure looks at a forecast that's short of plan and starts moving deals up a category — best case becomes commit, commit gets talked about in the room as good as closed — not because anything changed in the deal, but because the room needs a bigger number by Friday. The rep who flagged real risk on that account gets overruled in the room. Three weeks later the deal slips, and the post-mortem asks what the rep missed. Nobody asks who moved it.
We see a version of this constantly: a forecast call where a rep calls a deal "best case" with a specific, named risk attached — a champion who's gone quiet, a budget freeze not yet confirmed lifted — and a manager reclassifies it to commit anyway, because the stated risk is inconvenient for that week's number. The deal doesn't get less risky because someone changed its label. It just gets less visible. When it slips, it slips as a surprise, even though someone in the room saw it coming and said so out loud.
Both directions of this — reps sandbagging upward pressure, leaders inflating downward pressure — produce the same outcome: a forecast number that reflects what the org needs to hear more than what the pipeline actually supports. Software doesn't fix that. It just moves the lie into a cleaner field.
What CRM discipline can and can't buy you
InsightSquared's 2021 State of Sales Forecasting study, run with RevOps Squared across roughly 400 B2B organisations, found that when human bias and manual error are stripped out through automation, satisfaction with forecast accuracy climbs to 76%. Useful. But look at the causes that same research named for inaccurate forecasts in the first place: weak rep accountability, poor CRM data quality, and manual processes.
Notice what's missing from that list. Not "reps don't understand the product." Not "the market is unpredictable." The named causes are all about whether the number in the system reflects the truth of the deal — which is a discipline and incentive problem wearing a technology costume. You can buy forecasting software that removes rounding errors and duplicate entries. No software removes the incentive to round a shaky deal up before your manager sees it.
The test that tells you which one you've got
Pull your closed deals from the last two quarters. For each one, find out what forecast category it sat in 30, 60 and 90 days before the close or loss date — pull that from your CRM's stage history if it tracks changes over time, or from old forecast call notes and spreadsheets if it doesn't.
Two patterns to look for. First: how many deals marked "commit" at 60 days actually closed? If it's meaningfully below 80%, you don't have a data-hygiene problem, you have a category-inflation problem — deals are being called commit before they've earned it. Second: look at deals that jumped straight from early-stage or "best case" into "commit" in the final two weeks of a quarter, with no material change in the deal itself. That's not last-minute progress. That's the number being managed toward a target, not measured from reality.
Most leaders who run this test find both patterns, and they find them concentrated around quarter-end — the exact moment the forecast is under the most pressure to look good rather than be right.
When leadership is the actual forecast problem
Gartner's 2020 research projected that by 2025, more than 90% of B2B enterprise sales organisations would still be running forecasts on intuition rather than genuine data discipline. That prediction landed in the middle of a pandemic, long before this year's AI tooling wave — and yet Clari's 2025 data shows 87% of enterprises missing target regardless of record AI investment in exactly that gap. The tools got better. The number everyone's afraid to say out loud didn't move.
That's the tell. If better software were the fix, the accuracy numbers would already be improving in line with the spending. They aren't, because the thing producing inaccurate forecasts was never a tooling gap. It's what happens the moment an honest number reaches someone with the authority to be unhappy about it. If your reaction to a soft forecast is pressure rather than curiosity, you are training your team, quarter after quarter, to bring you a number that survives the meeting instead of a number that's true. They will comply. You'll keep being surprised.
Before your next forecast call, check this
Run these before you sit down to review the number, not after you've missed it again.
Ask what percentage of last quarter's "commit" deals actually closed, measured against your own historical data, not the industry benchmark. Ask whether any deal was reclassified upward in the final two weeks of a quarter with no new evidence behind the move — and who made that call. Ask your reps, privately, whether they've ever softened a risk they flagged because raising it created more friction than staying quiet. Ask what happens, specifically, to a rep who tells you the truth about a shaky deal three weeks before it slips — do they get thanked, or do they get pressed to "find a way." That last answer tells you almost everything.
The part nobody says in the forecast call
Here's the sentence that should replace "the forecast was wrong again" in your next leadership meeting: we built a process where telling us the truth costs more than telling us what we want to hear, and then we act surprised when we get told what we want to hear.
That sentence doesn't get said, because saying it means the fix starts with you — with what happens to the messenger, not with a new field in the CRM. A forecast is not a measurement. It's a relationship between what's actually happening in the pipeline and what's safe to report about it inside your building. Fix the safety, and the accuracy follows. Buy better software while the safety stays broken, and you'll just get a more precise version of the same wrong number.
If you want an independent read on whether your forecast problem is a data problem or an incentive problem, that's exactly what a proper sales audit uncovers. See how we run that audit, or talk to us before your next quarterly planning cycle.
Ready to hire or take the next step in your sales career?