Rich Get Richer: AI Is Widening the Gap Between Top Reps and Everyone Else

Generative AI arrived in sales with a promise that it would lift the middle of the bell curve: give every rep the research, drafting and call prep that only the best ones used to do, and the gap between stars and the rest would shrink. New research suggests the opposite is happening. The reps who were already winning are using AI to win by more, while average and struggling sellers are mostly using it to save a little time. For CROs whose number already depends on a small group of top performers, that is a concentration risk hiding inside an AI rollout.

The Gap Is Growing, Not Closing

The clearest evidence comes from a study reported by Inside AI on October 8, which combined interviews with eight C-level revenue executives from the Bain Capital Ventures CRO Advisory Board with a survey of 236 sales professionals at Stripe and Checkr. After AI adoption, 44% of top-quartile reps were assigned higher-value tasks, compared with just 12% of bottom-quartile reps. Managers were 16 percentage points more likely to report quality improvements for top performers than for the rest of the team.

The study's most useful finding is what drives the split. Top performers did report greater time savings, but the stronger factor was AI literacy. On a self-rated five-point scale, each one-point increase in literacy was associated with a 0.71-point increase in reported benefits. The best reps were not just handing AI their busywork. They were building their own tools around their existing strengths: at Checkr, one rep built a negotiation copilot before such tools were widely available, and at Brex an employee built an AI tool over a weekend that the company later rolled out more broadly.

The same pattern shows up at scale. Salesforce's State of Sales survey of 4,050 sales professionals across 22 countries found top performers are 1.7 times more likely than underperformers to use prospecting agents. They are also far more disciplined about the data those agents run on: 79% of high performers prioritize data hygiene, compared with 54% of underperformers.

Revenue Was Already Concentrated

A widening gap would matter less if performance were evenly spread to begin with. It is not. Salesloft's 2026 US Revenue Benchmark, a survey of 500 U.S. sales and revenue decision-makers, found the top 10% of sellers generate 47.4% of closed-won revenue while average quota attainment sits at about 62%. At the same time, 68.4% of leaders report higher pipeline quotas. Only 20.6% describe their AI strategy as production-ready with measurable outcomes.

Put those numbers together and the exposure is plain. If AI compounds the advantage of the reps who already carry half the number, the forecast becomes more dependent on a handful of people, and every resignation in that group costs more than it did a year ago. Meanwhile the majority of the team, the sellers whose improvement would actually move attainment, are getting the smallest return on the tools the company is paying for.

Support Teams Saw the Opposite Effect

None of this is an inevitable property of AI. When economists Erik Brynjolfsson, Danielle Li and Lindsey Raymond studied 5,179 customer support agents during the staggered rollout of a generative AI assistant, issues resolved per hour rose 14% on average, with a 34% improvement for novice and low-skilled workers and minimal impact on the most experienced. In support, AI narrowed the gap.

The difference is design. The support assistant was embedded in a defined workflow and suggested what the best agents would say, so newer agents absorbed expert behavior in every conversation. Most sales AI is deployed the other way round: a set of tools handed to reps, with the value left to whoever is skilled enough to find it. That model rewards the curious and the already capable. The Inside AI report describes the alternative its executives favor, a hybrid that centralizes the context layer, meaning data, permissions and guardrails, while letting reps discover and share tools from the bottom up.

Enablement is the other missing piece. Salesforce found 46% of sellers rarely get feedback on their sales conversations and 47% do not get enough roleplay before customer calls. Salesloft found only 56% of leaders say sellers are coached at least every two weeks. Accenture's Kevan Yalowitz made the point in the Inside AI report that A-players adapt as AI environments change, while others need structured training. Without it, the middle of the team never builds the literacy that the data says drives the return.

The CRO Playbook: Make AI Lift the Middle, Not Just the Top

• Measure AI impact by quartile. Report adoption, time saved and win-rate change separately for top, middle and bottom performers, so a rising average cannot hide a widening spread.
• Treat AI literacy as a trainable skill. Assess it, build it into onboarding and certification, and coach it on the same cadence as discovery and negotiation skills.
• Harvest what your top reps built. Find the prompts, copilots and workflows the best sellers made for themselves and turn the ones that work into shared team tools.
• Embed AI in the process, not beside it. Put guidance inside qualification, deal reviews and account plans so every rep gets the expert move at the moment it matters, the way support teams did.
• Centralize the context layer. Own the data, permissions and guardrails AI runs on, and close the data hygiene gap between high performers and everyone else.
• Watch your concentration risk. Track what share of closed-won revenue your top decile carries each quarter, and treat a rising figure as a forecast risk rather than a success story.

AI was supposed to be the great equalizer on the sales floor. So far it is behaving more like a multiplier for the people who were already ahead. That makes it less a technology decision than a design decision, and the CROs who build AI into the process and the training rather than handing it to reps as a perk will be the ones who get the middle of the team to move.