You have now sat in the meeting where the number gets committed something like twelve times.
Coverage is 3.5x. The forecast says the quarter lands. Everyone looked at the same slide and agreed, and you signed off, because the arithmetic worked and the people presenting it are not fools.
Then the quarter closed short, and there was an explanation. Deals slipped. A champion left. Procurement got involved late. The budget froze on their side. Every explanation was plausible. Several were true.
Twelve quarters. Twelve explanations. No pattern.
Here’s the thing you have never quite asked out loud, because there is never a good moment for it. Why is the explanation always retrospective? Nobody has ever walked into the forecast meeting and named the reason the number is going to miss. They name it afterward, fluently, every time.
That is not a competence problem. A room full of competent people cannot tell you in advance either.
What This Is Not About
Two things are true and neither one is the subject here.
Your attribution is imperfect. Three systems will give you three answers about where a deal came from, and none will match your CRM. You also have too many dashboards, and most requests for more detail are requests for reassurance rather than requests for a decision. I have written about that one already.
Both are arguments about the quality of the measurement.
This one is about whether the thing being measured is there at all.
The Process You Are Modeling Does Not Move in Stages
Your forecast assumes a buyer who enters at one end, moves forward through defined stages, and either converts or exits. Stage two to three at some rate, three to four at some rate. Multiply, sum, commit.
Gartner built the framework most B2B companies use to think about buying. Gartner also describes the actual process, in its own buyer enablement research, as „unpredictable, inconsistent and sometimes repetitive.” Their words. They compare the journey to „a disordered web rather than linear chevrons,” and found that 90% of buyers loop back to at least one of the six jobs they must complete before purchasing.
Sit with that for a second. The source your team quotes when they build the funnel model is the same source describing the underlying behavior as unpredictable.
And that research is from 2018, which means the era you think you lost was already gone before you started measuring it.
Gate an ebook, email everyone who downloads it, pass the list to sales, and you get something that looks like a machine: numbers big enough and a sequence visible enough to compute conversion rates and watch them move.
That is legibility. It has never been the same thing as control.
You did not lose your grip on the buyer. The buyer’s behavior simply became too visible to keep pretending the grip was there.
Everything Moved at Once, Which Is the Part Nobody Models
Say the process did behave like a funnel. A forecast still assumes the conversion rates you extrapolate from describe a system that is holding still.
It is not holding still. The Bridge Group’s 2026 research across 158 B2B companies found quota attainment down to 48% from 51% two years earlier, and ramp time at 6.2 months, the highest in the study’s history. Companies are buying more experience, 3.7 years on average against 2.7 in 2022, and getting worse results from it.
Then the part that matters most for your forecast. Compared with the previous year, near-majorities of those companies reported increases in stakeholder count, sales cycle length, discounting pressure, deal slippage, and required pipeline coverage.
All five, same direction, same time.
Their own conclusion is blunter than anything I would write: „a fundamental mismatch between market conditions and the operating model.”
A model fitted to last year’s conversion rates is fitted to a process that no longer runs.
So Where Does the Confidence Come From?
In 1975, a Yale psychologist named Ellen Langer published a paper on something she called the illusion of control, which she defined as „an expectancy of a personal success probability inappropriately higher than the objective probability would warrant.”
Take a situation governed mostly by chance, add factors that belong to situations governed by skill, and people become confident out of all proportion to the odds. The factors she tested were competition, choice, familiarity, and involvement. Her summary: „the more similar a chance situation is to a skill situation, the more likely it is that people approach the chance situation with a skill orientation.”
Now look at what a pipeline review consists of. Stages somebody chose. Probability weights somebody set. A weekly rhythm that keeps everyone involved. Named accounts everyone recognizes. Competitive benchmarks in the corner of the slide.
Langer’s list, one item at a time, rebuilt as software.
The confidence in that room is manufactured by the instrument. It is not extracted from the odds.
Langer also explains why nobody wants to give it up, and that is the part I did not expect to find. People chase control to avoid „the negative consequences that accompany the perception of having no control.” Her line is that „a temporary loss of control is anxiety arousing,” and that a false sense of control over a coming event genuinely reduces how threatening it feels.
So the quarterly forecast is doing a job. Just not the job on the label. It is an anxiety instrument, and it works at that whether or not it is accurate, which is why demand for it survives twelve consecutive misses.
She is honest about the limit, and so should we be: „there is an element of chance in every skill situation and an element of skill in almost every chance situation.” Selling is not roulette. Nobody is claiming it is.
The Part That Should Bother You More Than It Does
I went looking for a credible measurement of how accurate B2B sales forecasts actually are. Old habit from my years in trade journalism, where you do not print a number until you have found the person who produced it.
I could not find one.
Every figure in circulation traces back to a company selling forecasting software, or to a citation with no report attached. „79% of sales organizations miss their forecast by more than 10%” is quoted everywhere and credited to a research firm that appears never to have published it.
Then it gets worse. The most-quoted win rate statistic in B2B sales, that win rates have fallen to 19% from 29%, comes from a benchmark study of 655,000 opportunities that never published an absolute win rate. The study reports relative year-over-year change and says so plainly: „All percentage figures are relative.” The real numbers are minus 18% one year and minus 10% the next, describing a decline that got shallower. Somebody read a rate of change as a level, reversed the finding, and the field repeated it.
The same report carries a guarantee from its publisher. Forecast accuracy to plus or minus 10%, within six months.
An industry that sells predictability cannot produce a sourceable measurement of how predictable anything is, and misreads its own most-quoted dataset.
Where Predictability Is Real
It would be dishonest to write all of that and not say where the opposite holds.
Predictability scales with deal count. A company closing several thousand small self-serve transactions a quarter can forecast, genuinely and well, because large numbers do the work no dashboard can do. The noise cancels out.
You are probably not running that company.
Close forty significant deals a year and one slipping moves your number by 2.5%, five by 12.5%. Run a software development firm or a technology services business where a dozen engagements make the year, and a single delayed statement of work swings a quarter on its own.
That is a small-numbers process being managed as though it were a system, and no amount of reporting resolution changes the arithmetic.
What the Chase Costs You
Here is where it stops being philosophical.
Chasing a number that cannot exist buys you reporting. In a study of 750 senior marketing leaders, commissioned by a demand generation vendor and cutting against their own category, two thirds said their dashboards regularly show success that does not translate into revenue. 85% said their teams spend more than half their time fixing things rather than building anything.
AI made the reporting nearly free. A weekly deck that took two days now takes minutes, so it happens weekly, then twice weekly, and no single request is unreasonable.
What got displaced was the work nobody can put on a slide.
Finding out why the last five deals were actually lost, from the buyers rather than the reps. Rewriting the offer because the market moved. Saying one thing your buyer recognizes as true about their own situation.
None of it produces a number by Friday. All of it moves whether you get chosen.
The Question Worth Asking Instead
Unpredictable is not the same as uninfluenceable. That distinction is the whole thing.
You cannot engineer when revenue arrives, because the timing belongs to a buyer whose internal circumstances you will never see. What you decide is whether the work you paid for this quarter is still working in the next one, or whether it evaporates and the function starts clean again in October. That is what structure buys you, and it is the variable you actually hold.
So before the next forecast meeting, ask for something different. The last eight quarters of forecasts, next to what actually landed.
If the error is large and leans the same way every time, you are not looking at execution failures. You are looking at an instrument with a bias, and reading a biased instrument more often does not make it more accurate. It makes it more expensive.
Then ask the better question. Not what next quarter holds, because nobody can tell you that, including the people selling you the tools that promise otherwise.
Ask what your company built last quarter that is still earning this one.
You control the answer to that one completely.
