The Sales Forecasting Problem Nobody Wants to Talk About

Every sales forecasting process I've ever seen has a moment like this one: Picture a regional sales director at a mid-market software company. I'll call her Diana. Tomorrow morning she owns the number in her VP's Q3 forecast review, and right now it's past ten on a Sunday night and she's still at the kitchen table with her laptop open. In front of her: a slew of dashboards that slice pipeline data every way imaginable and forty-odd CRM notes from her reps, most of which amount to "feeling good about this one." Her new AI forecasting tool has weighed in too, with a number backed by confident reasoning that lands nowhere near her own math.

She has more information than any sales leader a generation ago could have imagined, and somehow, she's less confident than ever. Like all of us, she'd hoped that AI would be the silver bullet to this decades-old problem, and while it has that potential, there's one fatal flaw: its reasoning rests on the same shaky and biased data she's struggling to make sense of. And she's far from alone. After another two hours trying to frame all of it into a picture she believes, she does what she does every quarter: She goes with her gut.

Sound familiar?

I've spent more than twenty years around forecast reviews, and unfortunately, Diana's night is the rule, not the exception. Over the years, we've bought CRMs, revenue intelligence tools, and now AI stacked on top of all of it, and most sales forecasting still amounts to fancy guesswork. In fact, in Xactly's 2024 forecasting benchmark, four in five leaders admitted to missing a quarterly forecast in the past year, and more than half missed it at least twice. That's why I want to dig a little deeper in today's post, to the root of the problem: a lack of a shared language and objective criteria. Then, I'll detail actionable strategies leaders and reps can use to improve forecast accuracy without disrupting daily workflows or buying extra tools.

More CRM Data Hasn't Made Sales Forecasting More Accurate

Most forecasting failures are failures of language before they are failures of analysis.

For example, walk any deal from the AE who owns it up to the CRO, and watch what happens to the information along the way. The AE checks the boxes and calls an opportunity "strong." The manager reads "strong" and rolls it into the commit. The VP hears that and passes it up the chain. Nobody lied. Everyone made what they thought was the right call. And yet, every person in that chain used the same word to mean something different, and the number that reached the top was built on a shaky foundation.

That's the organizational game of telephone running underneath most forecasts. No tool can fix it, because the data in your system exists through the filter of each rep's data entry. If ten reps define "qualified" ten different ways, a more expensive platform gives you their inconsistencies faster and in better-looking charts. It's a problem that's more widespread than most realize — according to research by Validity, 76% say less than half of their CRM data is accurate and complete.

Forecast Accuracy Breaks Down in Translation

The reason the language is so slippery is that most organizations have never agreed on what good looks like and what the terms used to discuss opportunities actually mean.

Ask five reps what makes an opportunity qualified and you'll get five nuanced answers. They might all fill out their qualification tool the same way, but one leans more heavily on budget, another points to a champion, and a third hangs their hat on a demo that landed well. Each is describing a different part of the same picture, and none of them is describing it the way the person sitting next to them would. When that's true, forecast accuracy is not something you can coach your way to one deal at a time because the vocabulary itself has broken down.

This is why I've argued for years that qualification is a continuous conversation, guided by objective criteria. When it gets treated as a single box a rep ticks on the way to the next stage, that's where so many forecasts quietly go wrong.

Why Sales Qualification Decides the Forecast

Most teams already run a qualification tool: MEDDPICC, BANT, or a hybrid built over the years. And while a rep's pipeline shows every box checked — metric captured, economic buyer named, pain identified — roughly a third of those deals still get stamped no-decision by quarter's end.

What falls by the wayside is the harder conversation:

• What business issue is actually driving this purchase?
• Who else has to say yes?
• What are the problems worth solving here?
• Has the value of solving these problems been quantified, and does the buyer believe that number?
• What happens to this buyer personally if they solve the issues, and what do they stand to lose if they stick with the status quo?

The answers to those questions are what tell you whether a deal is real, and most status fields can't hold them, so they never make it into the number Diana carries into the review. That's why sales teams must ensure mechanisms are in place to capture and evaluate this information on a continual basis.

A Shared Language Changes What a Forecast Is Worth

Give a revenue team one common language for describing and evaluating opportunities, and the forecast stops being one person's gut feeling. It becomes a shared judgment that holds up under questioning.

I watched this happen at a global engineering firm we worked with that routinely competes against companies five times its size. It couldn't afford to pour resources into proposals for deals it was never going to win, so the whole organization — sales, engineering, finance, and legal — learned to qualify opportunities against the same criteria and describe them in the same words. As a result, average deal size rose 833% and cost of acquisition fell 40%.

Yes, the numbers are incredible, and I chose this outlier of an example for two reasons: First, it shows how much potential could be hiding in plain sight. Second, while those revenue numbers are every sales leader's dream, the cultural shift underneath them mattered more. When everyone started evaluating deals the same way, forecast reviews became a real conversation rather than a defense; people asked sharper questions; reps were more willing to point out weak spots; knowledge sharing improved; and ultimately, putting in the tremendous amount of work required to pursue these complex opportunities became more efficient and more satisfying.

I realize that some may have been skeptical of that last point. The notion of culture can feel ephemeral, especially inside the sales organization. However, it has real consequences. For example, take a second to sit with this: In Validity's 2025 CRM data study, 37% of employees admitted they regularly fabricate data to tell leaders what they want to hear. And if 37% say they're "fabricating" data, what about the other 63%? Would it be unreasonable to think that while they're not fabricating data, they might be entering it through rose-colored glasses?

After all, we hire salespeople because they're good at reading the person across from them and adapting their behavior accordingly. Plus, people want to do good work, and they gravitate toward what they believe is expected of them. So, before you challenge your reps' numbers, look hard at what your culture actually rewards. If candor about a stalled deal earns a public grilling, and a green dashboard earns a nod, your team already knows which one to bring you. In the end, forecast accuracy is downstream of the signals a leader sends.

Building a Sales Forecast You Can Trust

Let's check back in on Diana again at the kitchen table, past midnight, surrounded by data she couldn't trust. Everyone around her would assume her problem was analytical, that with a better model or a cleaner dashboard the picture would finally resolve. However, now we know that she wasn't facing a math problem, but a team that described the same deals in a different language, and a stack of CRM notes that captured their optimism instead of their reasoning.

This is why more data, and now more AI, hasn't fixed forecasts. A model can only be as reliable as the human judgment feeding it, and a confident answer built on inconsistent inputs is a faster route to the wrong number. The work that actually moves a forecast is unglamorous and hard: getting every person on the revenue team to describe, qualify, and evaluate a deal the same way, in the same words, against the same criteria.

That's the forecast problem nobody wants to talk about. You can keep buying your way to more data and still end up exactly where Diana did, alone at the kitchen table with a call she can't defend, or you can opt for a reliable approach to forecasting based on a shared language and common approach. In my experience, it's one of the most durable investments a sales leader can make.