I’ve worked in and around a lot of companies now, and I’m not sure I’ve ever seen revenue forecasting genuinely work well. This is part one of two. Today, what I think is actually broken. Later this week, a model I’ve been working on to fix it.
Let me start with a confession.
I’ve sat in more than one quarterly business review where everyone in the room knew the number was soft and nobody said it out loud. I’ve presented pipeline coverage I didn’t fully believe, because the ratio looked fine and questioning it would have opened a conversation nobody had time for. And I’ve forecasted marketing-sourced pipeline and missed it, more than once, and told myself the deals slipped rather than admitting they were never real.
I don’t think I’m unusual in that.
Gartner’s State of Sales Operations research found that only 45 percent of sales leaders and sellers have high confidence in their organization’s forecasting accuracy. Not the board’s confidence in the forecast. Their own. The same research found that just 7 percent of sales organizations achieve forecast accuracy of 90 percent or better.
If the people closest to the deals don’t believe the number, this isn’t a reporting failure or a discipline failure. Something structural is wrong with how we build our revenue forecasts to be reliable.
And the damage compounds. Boards understand variance; markets move and deals slip. What ends tenures, in my experience, is missing the number after telling the board the last miss wouldn’t repeat. That’s when it stops being a performance conversation and becomes a credibility conversation, and I’ve rarely seen folks recover from the second one cleanly.
Why I’ve Stopped Trusting Pipeline Coverage
Coverage is the most quoted number in go-to-market forecasts and I’ve come to trust it the least. Four reasons, and none of them require a benchmark to see.
When raw coverage treats every dollar as equal. If an early-stage opportunity and a late-stage opportunity of the same size contribute identically to the ratio, and obviously not equally likely to close, the forecast will not be reliable. Two companies can report the same coverage and face completely different quarters.
Stale deals inflate it. Unqualified opportunities, deals that should have been closed out months ago, prior-period slips. All of it sits in the ratio making the number look healthy.
Phantom pipeline is the one that really gets people. A late-stage deal slips from Q1 to Q2, then Q2 to Q3, carrying its full value into the coverage ratio each time. Counted three times. Never real once. I’ve watched this happen and not caught it until the third quarter.
And then the quiet one: three functions, three formulas. Finance calculates coverage one way, sales another, marketing a third. Nobody is exactly wrong, and the planning conversation breaks before it starts because everyone is defending a different number. I think that’s the Silo Tax showing up in the forecast, and I’d argue it costs more than most campaign decisions anyone will make this year.
My Argument: The Damage Happens Upstream
Here’s where I’d push back on how most forecasting conversations get framed. They start at the pipeline and work forward. I think most of the damage was done before anything entered the funnel.
Matt Dixon and Ted McKenna made this concrete in “The JOLT Effect.” Using Tethr, the AI conversation intelligence platform where Dixon serves as chief product and research officer, they analyzed more than 2.5 million recorded sales conversations and found that somewhere between 40 and 60 percent of deals end in no decision at all. Not lost to a competitor. No decision. When they looked at why, 56 percent traced to customer indecision driven by fear of getting it wrong (FOMU – Fear of Messing Up), and 44 percent to a preference for the status quo (Not buying into FOMO – Fear of Missing Out).
If that’s right, then a large share of what teams forecast against was never a real deal. It was a conversation that hadn’t failed yet.
Six things seem to determine whether pipeline is real, and not one of them lives inside the forecast process.
At-bats. How many deals you’re invited into at all. This is the one I think gets left out most often, and it may matter more than everything else on this list, because it’s multiplicative. Win rate only matters against deals you’re actually in.
Kerry Cunningham spent years as a Forrester analyst (as did I) before joining 6sense, where he runs an annual study of B2B buyer behavior that I’d argue is the most useful research in our field right now. The 2025 edition surveyed nearly 4,000 buyers across North America, EMEA, and APAC, with a median purchase value between $200,000 and $300,000.
Buyers assemble their shortlists by consuming content on day one, before contacting anyone, filling roughly three and a half of five slots of vendors they will consider. In 2025, 95 percent of the time the vendor that eventually won was already on that day one list, up from 85 percent the prior year. Buyers then rank that list by preference before speaking to a single seller; 94 percent of buying groups said they had done exactly that. They initiate contact themselves 79 percent of the time, reaching out to their preferred vendor first. And the first vendor contacted wins about 80 percent of the time.
The obvious objection is that first conversations are simply persuasive. Cunningham’s team tested that. In the roughly six percent of cases where buyers had not pre-ranked their shortlist, the first vendor contacted won only 57 percent of the time, not more. Being contacted first turns out to be evidence that you were already preferred, not the cause of it.
If nearly all winners were on a short, pre-ranked list assembled before your sales team knew the deal existed, then most of your quarter was decided before a single opportunity was created in your CRM.
The same study found that 85 percent of buyers had prior experience with the vendor they ultimately bought from, 97 percent had prior experience with at least one vendor on their shortlist, and 70 percent said economic uncertainty pushed them toward more conservative choices, meaning known vendors and incumbents. Being known isn’t an advantage in that environment. It’s closer to a prerequisite.
This is what I’ve been calling Brand Gravity, and I’d argue it’s the least understood input to forecast accuracy in B2B. It also reframes the SiriusDecisions B-to-B Buying Study finding that 89 percent of C-level executives say their perception of a provider’s brand reputation influences shortlist creation moderately or significantly. That isn’t a just a brand statistic. It’s a pipeline statistic as brand reputation strongly seeds demand, albeit with a lag not always easy to forecast directly.
You can’t forecast your way onto a list you were never on.
Brand and Demand quality. My working belief is that pipeline from earned demand behaves differently than pipeline that was bought. Converts higher, slips less, holds price. I’ll be honest that I haven’t seen a study measuring pipeline behavior by demand source, and I’d like to.
What I can point to is that Fred Reichheld , the creator of Net Promoter Score from Bain & Company, built an entire metric on this distinction. In Net Promoter 3.0, written with Darci Darnell and Maureen Burns for Harvard Business Review in 2021, Reichheld introduced the earned growth rate, which captures revenue growth from returning customers and their referrals. His test for classifying a customer is refreshingly blunt:
Ask every new customer why they came on board. If the reason is a referral or a recommendation, that customer was earned. If it was advertising, a promotional deal, or a persuasive salesperson, that customer was bought.
Reichheld created it because self-reported NPS had been gamed so widely it stopped meaning anything to investors. Earned growth draws on accounting data instead, which makes it auditable. It’s a customer loyalty measure a CFO can actually sign. I’d argue the same logic applies one step earlier, in the pipeline. If you can classify closed customers as earned or bought, you can classify open pipeline the same way. I don’t know anyone doing that systematically. I think someone should.
ICP fit. I’d guess bad-fit accounts are among the largest sources of the no-decision problem Dixon and McKenna documented. They engage, take meetings, build an internal business case that doesn’t survive their own CFO, and disappear. Every one of them sat in somebody’s forecast.
Win rate. A few points of movement here changes required coverage more than almost any top-of-funnel effort, and it costs nothing incremental. I’d rather fund win rate than volume in most portfolio companies I see.
Deal size stability and pricing power. If average deal size swings meaningfully quarter to quarter, I don’t think any coverage ratio can produce a reliable forecast, and that’s usually segmentation or packaging rather than selling. Related: deals that require heavy discounting to close probably weren’t won on value. I suspect discount depth is a leading indicator of forecast slip and I rarely see anyone track it that way.
Marketing has meaningful influence over four of those six and shared ownership of all of them. Which is why I don’t think this is a sales problem, and I’d be curious whether you agree.
The Bottom Line
Only 45 percent of sales leaders have high confidence in their own forecast. I’m not surprised the board doesn’t either.
My argument is that the forecast isn’t broken at the forecast. It breaks upstream, where brand reputation determines at-bats and 95 percent of winners were already on a list drawn before anyone in sales knew the deal existed. It breaks in the middle, where coverage ratios count phantom deals three times. And it breaks across functions, where three teams compute three numbers and call the disagreement a data problem.
That’s the diagnosis, and I’m reasonably confident about it.
Later this week, I’ll put up part two, which is a model I’ve been working on for actually fixing this. It’s called Pipeline Triangulation, it needs three functions in the room to work, and I’m less certain about it than I am about the diagnosis so I welcome you to respond and argue with it.
Outcomes over activities. Every time.
Sources
Gartner, State of Sales Operations Survey, on forecast confidence and accuracy.
Kerry Cunningham and 6sense, 2025 B2B Buyer Experience Report, approximately 4,000 buyer responses across North America, EMEA, and APAC, median purchase value $200,000 to $300,000.
Matthew Dixon and Ted McKenna, The JOLT Effect, based on analysis of more than 2.5 million recorded sales conversations using the Tethr platform.
Fred Reichheld, Darci Darnell, and Maureen Burns, “Net Promoter 3.0,” Harvard Business Review, November-December 2021.
SiriusDecisions, B-to-B Buying Study, on brand influence in shortlist creation.
