By now, nearly every sales organization has brought AI in the door. Salesforce’s State of Sales report puts it at 81% — teams either experimenting with AI or already running it in production, across prospecting, forecasting, drafting, and call summaries. That’s good news; the tools are useful today, and they’ll do more tomorrow.
The question worth a sales leader’s attention is what that “more” looks like — and whether any of it has actually made your people better at selling. That one is harder, and for most of the history of sales it wasn’t a question anyone could answer.
The Half That Was Never Visible
Most of what teams adopted this year is, in effect, a brilliant intern. It drafts the follow-up, cleans up the call notes, assembles the sequence — capably and fast. Keep it; it’s genuinely useful. But an intern has never carried a quota, doesn’t know your methodology, and can’t tell you whether a rep is improving or just staying busy. It helps with the activity, not the skill.
That distinction is the whole game, because sales performance has always been the product of two things — Activity × Proficiency — and only one of them was ever visible. Activity you can count and manage. Proficiency you couldn’t. Companies have spent millions on training in the hope of lifting it, but once the workshop ended, no one could measure whether skill actually moved, seller by seller, or rebuild it reliably in the field. So the more powerful half of the equation sat largely untapped — not because it didn’t matter, but because it couldn’t be seen.
What It Takes to Build the Skill
So how do you build the half that’s been invisible? The same way any skill has ever been built — through practice measured against real expertise. Vince Lombardi put it simply: practice doesn’t make perfect; only perfect practice makes perfect. The psychologist Anders Ericsson spent a career establishing why — deliberate practice, focused repetition with immediate feedback against a clear model of excellent performance, is what turns effort into mastery, while unguided repetition just plateaus. A golfer alone on the range grooves whatever swing they bring; the same golfer with a coach who knows exactly what great ball-striking looks like gets measurably better. The reps are identical. The expertise evaluating them is everything.
That is what most “AI for sales” is missing — and it’s where the common assumption goes wrong. People imagine the hard part is the methodology: that once you have one, you teach it to the AI and you’re done. It’s very nearly the reverse. A frontier AI is already intelligent; you don’t make it an expert by handing it your playbook to read once. You make it an expert by putting your entire way of selling within its reach the moment it reasons — every conversation and every stage, the models as they’re actually executed, worked examples from prospecting through objection-handling, all instantiated in your real environment, against a clear standard for what great looks like. The intelligence was never the scarce ingredient. That organized body of expertise, on hand the instant it’s needed, is.
So genuinely developing a seller takes far more than a passing familiarity with your methodology. It takes an AI that is expert in three distinct ways.
First, expert in your methodology as it operates in your business: your buyers and what they value, your competitors and how they show up, and what an excellent conversation sounds like when your rep runs your methodology on your kind of deal — not a generic script, but the real thing, in your market.
Second, expert across every conversation and scenario your people actually face — the full range of situations a seller walks into, not a handful of tidy demonstrations.
Third, and this is the one most easily underestimated, expert at evaluating execution. It needs a standard for what good looks like that is more discerning than the sellers being measured against it — sharp enough to tell whether a rep is running the methodology with genuine command or simply reciting the words they were handed. On paper those look alike; in front of a buyer they are worlds apart. And it has to turn that read into feedback the seller can act on — specific, and pointed at the next thing to work on.
What Becomes Possible
Build AI to that standard and something genuinely new happens: the invisible half of the equation becomes visible. Proficiency stops being a hunch and becomes a measurement — you can see where a seller is strong, where they’re slipping, and whether they are actually getting better.
And once you can measure it, you can move it. Consider the kind of thing that simply couldn’t be produced a few years ago: across the sellers practicing this way, roughly 65% demonstrate proficiency on a new skill — and when a seller practices a skill more than once, their score improves by about 48% on average. The point isn’t the particular figures; it’s that they can be produced at all. The half of sales performance that stayed dark for decades is now something you can put a number on, watch move, and coach against.
But the real power isn’t in the aggregate. Those figures are rollups of something far more useful: a specific, per-seller reading of proficiency — not a randomized survey’s estimate of how the team is trending, but a clear picture of which sellers, on which teams, are fully proficient on which skills, and at what level. Proficiency isn’t one thing you either have or you don’t; there are degrees of it, and they can finally be told apart.
That is what turns measurement into management. A manager knows exactly what each seller is genuinely good at — and can set the expectation that they bring it to the field. They can see who is still developing and who has mastered a skill, aim their limited coaching time at the person and the moment that need it most, and move each rep forward from where they actually are, rather than from where a training roster assumes they should be. Coaching stops resting on instinct and the memory of the last ride-along, and starts resting on evidence — because now the skill, and its progress, are visible.
The Question Worth Asking
So as you weigh the AI arriving on your sales team, the question isn’t how clever the tool is, or how much it can generate. It’s whether it understands what great selling looks like in your world — deeply enough to tell whether your people are getting better, and to help them do it.
Your brilliant intern is worth keeping. It was just never going to coach your team. That takes an expert.
At Axiom Sales Kinetics, this is the work of three decades: a selling methodology proven in the field, now operating as a true expert inside each client’s own environment — so that proficiency, the half no one could ever measure, is finally visible, coachable, and improving. If you’d like to see what that feels like firsthand, walk through a Skill Path — about fifteen minutes, no account required: axiomsaleskinetics.com/skill-path.
