The human-AI balance in B2B sales: What buyers actually want and how to deliver it

Editor's note: This article was based on a talk given by Rob Perrilleon, SVP, Client Experience at Corporate Visions, at our Sales Enablement Summit, San Francisco, 2026.

There's a statistic that caused a lot of panic a couple of years ago. Gartner put it out, and it spread fast through sales and enablement circles. Seventy-five percent of buyers, they said, prefer a rep-free buying experience. No humans. Just digital, self-serve, automated.

A lot of people took that seriously. Some started questioning whether the entire profession of sales enablement had a future.

Then, earlier in 2026, Gartner published another finding. Same number, different story. Seventy-five percent of buyers who had experienced that rep-free, AI-driven buying journey said, on reflection, they'd actually like some human interaction at the critical moments. The same proportion of buyers who supposedly didn't want humans had gone through the experience and changed their minds.

So what does that tell us? It tells us the question was never really about humans versus AI. It was always about the right mix. And figuring out that mix, specifically what human interactions still drive wins, how you assess who's actually good at them, and how you scale those capabilities with AI, is one of the most important things revenue and enablement teams can be working on right now.

What human sellers actually contribute to wins

When we work with B2B companies, one of the first things we look at is the gap between how sellers explain their wins and losses versus how buyers actually describe their decisions. The difference is significant.

We analyzed 150,000 B2B purchase decisions through buyer interviews, and the findings were clear: seller-reported reasons for winning or losing a deal matched the buyer's perspective only about 30% of the time. The other 70% of the time, sellers and buyers were describing the same decision in completely different terms.

Here's how that typically plays out. When a seller wins, they tend to attribute it to their product, their relationships, or their own performance in the sales process. When they lose, it's usually one of three things: the product lacked the right functionality, the price wasn't competitive, or the buying committee politics made it impossible. Three factors that feel largely out of their control.

Buyers tell a different story. When we looked at the aggregate data, buyers reported that in more than half of the deals where they chose a competitor, the losing seller could have turned that into a win. And the thing that would have made the difference was within the seller's control.

That's a massive opportunity sitting right there. Preventable losses that aren't being prevented, because sellers are attributing outcomes to factors they can't influence rather than examining the behaviors they can.

So we did the analysis. We looked at which specific buyer experiences and seller behaviors were actually predictive of wins across that dataset. The result is a competency model that we're sharing openly with the industry, because we think it's more useful out in the world than sitting in our hands.

A few important things about this model. First, every behavior in it is predictive of higher win rates. We calculated correlation coefficients for each competency, and anything above 0.5 would be considered non-predictive.

Everything in this model sits materially above that threshold. Second, and this matters just as much, every behavior is observable, teachable, and coachable. These aren't vague personality traits. They're things you can actually assess and enable against.

We also looked at whether the winning behaviors differ between new logo acquisition and account expansion. They do, in meaningful ways. There's some overlap; responsiveness and quality of communication matter in both contexts, but the competencies that drive wins in a new business conversation are largely different from those that drive retention and expansion. That distinction has real implications for how you structure your enablement programs.

Why your win/loss data might be misleading you

Most organizations are still relying on CRM fields as their primary source of win/loss insight. Sellers self-report, and those reports go into the system. It's a data point. It's a signal. But given what we've seen about the gap between seller perception and buyer reality, it's worth asking how much weight you're putting on it.

The customer's point of view is the missing piece in most win/loss analyses. When you bring that in, through structured buyer interviews or aggregate purchase decision data, you often find a very different picture of what's actually moving the needle.

The practical implication for RevOps teams is to think carefully about what data sources you're combining when you're trying to understand why deals are won or lost. CRM data tells you what sellers think happened. Conversation intelligence gives you what was actually said. Buyer feedback tells you what actually mattered to the decision. All three together give you something much closer to the truth.

Assessing who's actually good at what

Once you know which behaviors drive wins, the next challenge is figuring out who on your team is good at them and who has gaps. That's the assessment question, and it's where technology is moving fast.

We've been working with an expert who came out of the Harvard Skills Lab, which focuses on assessing what they call complex skill clusters. Sales is one of those. There's a lot happening at once in a sales conversation, and isolating what's actually driving outcomes is harder than it looks.

The criteria for a useful assessment in this context are fairly specific. First, it has to be realistic and in context. Sales is messy. You don't get accurate data from a micro-assessment that isolates one skill in an artificial scenario, because in real selling, behaviors blend together. There's no such thing as a call where you're only doing one thing.