Everyone shopping for sales AI this year is asking whether the technology is ready.
Far fewer are asking whether they are.
That's an easy question to skip, because it doesn't feel like it belongs in a software evaluation. And for most of the last decade it didn't. But the moment you point an AI at seller development, your methodology stops being a training asset and becomes an input — and inputs have requirements that training assets never had.
We've made the case before that when no one defines good selling, your AI will. That piece was about not having a standard at all. This one is about having one your AI can't use — which is harder to catch, because from the outside it looks solved.
You've Seen a Methodology Go Unused. This Isn't That.
You already know what an unused methodology looks like. It gets defined, rolled out, celebrated at kickoff — and then quietly not used, while sellers go on selling the way they always have. It isn't rare. In CSO Insights' 2019 sales enablement study, only 34.7% of organizations could say that more than three quarters of their sales force used their defined sales process daily, and the researchers noted the same pattern held for methodology adoption.
That study is a few years old, and nobody has argued since that it was too pessimistic. Anyone who has run a sales team recognizes the number.
That failure was never really about the methodology. It was about reinforcement — one-and-done training, coaching that didn't happen, nothing measured after the workshop ended.
The AI version looks identical from the outside and has nothing to do with any of that. Your AI isn't skipping the hard step because the quarter closes Friday. It hasn't forgotten what the workshop covered on handling objections — forgetting isn't something it does. It is ignoring your methodology because it cannot read it.
Same symptom, entirely different cause — and everything we learned about fixing the first does nothing at all for the second.
Methodology Isn't Binary
Most conversations treat this as a binary state. You have a methodology or you don't.
There are three states.
You don't have one. Nobody has defined what good looks like, so there's nothing to be ready with. Uncomfortable, but at least it's visible.
You have one, and nothing can read it. It was defined, documented and rolled out. It lives in a deck, a workbook, an LMS module, and the judgment of your three best managers.
You have one in a form a machine can actually use. Structured, specific, and reachable at the moment the AI does anything.
The middle state is the one worth talking about. It's where the effort went. And from the outside — from the vendor demo, from the pilot, from the first quarter of usage — it is indistinguishable from the third.
Written Down Is Not the Same as Readable
Think about where your methodology actually lives right now.
There's the deck from the rollout. The workbook people got in the session. A few recorded modules. Some battlecards that were current two product launches ago. And the largest share of it sits in the heads of the managers who have been running this play since before the last reorg.
All of that is real. Almost none of it is reachable.
So when someone points an AI at a seller and asks what good looks like here, the AI does not stop and tell you it can't find your standard. It answers. It blends what it learned from everything ever written about selling with whatever fragments of your own material it happens to have been handed, and it returns something fluent, structured and complete.
And this is the part that makes it hard to catch.
It sounds like you.
A generic answer is easy to spot — anyone who knows the business reads two lines and rolls their eyes. An answer that has picked up your product names, your segment language and a couple of your objections reads as though the system understands your business. It doesn't. It has learned your vocabulary, which is a very different thing from learning your judgment.
That's not a failure of anyone's effort, and it's not a knock on the tool. Nobody built their methodology for a machine to read, because until about two years ago there was no reason on earth to.
The First Test: Go Find It
Pick something basic. How your sellers are supposed to uncover what a buyer will actually decide on — not the stage in your process where that's meant to happen, but the thing the seller does.
Two questions about it. Is there a named model for it? And are there examples of what it sounds like in your business, with your buyers, about what you sell?
Go find them. Two minutes, and you can't ask anyone.
Most people don't find both. They find a deck that references the idea, a module that teaches around it, and a manager who could explain it perfectly in thirty seconds if they weren't in a QBR.
If you can't retrieve it in two minutes, neither can a machine. And the machine doesn't have the option of walking down the hall.
The Second Test: Could It Answer, or Only Read?
You can hand your AI the workbook. It will read it — all of it, more carefully than most new hires do.
That still doesn't make it useful in the moment, and the reason is worth being precise about.
Nobody asks a well-formed question in the middle of a deal. Your seller doesn't say "retrieve the decision criteria model." They say "she keeps telling me they're happy with who they've got." Nothing in that sentence names a skill.
So before the AI can be any use at all, it has to work out what that moment actually is, pull the model that applies, find the example of what it sounds like with your buyers, factor in what this particular seller tends to get wrong — and assemble all of that into one coherent response, while the conversation is still going.
Not a list of relevant documents. An answer.
A workbook can't support that, and the reasons are unglamorous. The material has to be text, not pictures of text — a slide is an image of an idea, not the idea. It has to be broken into pieces small enough to point at: that model, not chapter three. And the pieces have to be joined by connections that say what kind they are — this lesson teaches that skill, this conversation demonstrates it, this assessment measures it. Knowing two things are related isn't enough. A machine needs to know how they're related to know what to do with them.
None of that is exotic. It's just work that nobody had a reason to do until recently.
It's a Chain, Not a Score
Here's the part that decides where to spend the next dollar.
These don't add up. They chain — and the weakest one sets your ceiling.
Perfect connections across models nobody ever defined buys you nothing. Perfect definitions with nothing joining them buys a lookup tool, not a coach. Beautifully structured content with no examples from your own business gives you a machine that can recite your methodology and has never heard it spoken.
So the question isn't how many of these you have. It's which one is weakest — because that's the one your AI is actually operating at.
A Better Model Will Not Fix This
The most common response we hear is that this is a temporary problem. Models are improving quickly; surely one of them will eventually understand how your company sells.
They won't, and it's worth being precise about why.
Training is what a model already learned — everything ever written about selling. It's genuinely useful and it's improving fast. It is also identical in kind for everyone, and whoever has the best one today, your competitor can buy the same one tomorrow.
Grounding is different. It's what the AI can reach and actually use at the moment it acts: your methodology, your environment, and over time what it learns about each of your people. Nobody ships that in a model release, because none of it exists anywhere except in your business.
Which means the better the models get, the more the difference between two sales organizations comes down to what their AI can read. The technology gap closes. The grounding gap doesn't — unless somebody closes it deliberately.
Start With One Skill, Not With a Project
If both tests went badly, the instinct is to scope a codification project — get everyone in a room, work out what the whole methodology should be, come back in six months. That instinct is why this rarely gets done.
Pick one skill instead. The one your managers coach most often, or the one your best sellers visibly do differently from everybody else.
Define what a seller actually does and says, step by step. Write two examples from your own business — real buyers, real language, the way it actually goes here. Then connect it to whatever teaches it and whatever measures it, so those three things know about each other.
That's a week or two of somebody's attention, not a program. And it does something no planning exercise can: it tells you honestly what the rest of the job would cost, because you've now done a representative slice of it.
One skill done properly beats forty done loosely. It's also the only version of this that survives contact with a quarter.
The Work Is Boring, and It Compounds
None of this argues for slowing down on AI. It argues for being honest about the order of operations.
Defining a skill at the behavioral level, building worked examples in your own environment, structuring it so something other than a human can retrieve it — that is unglamorous work, and it is the highest-leverage thing most sales organizations could do this year. Not because it makes the AI smarter. Because it is the only part of your AI that your competitors cannot also buy.
Activity × Proficiency = Sales. AI applied to the left side of that equation gets faster at whatever you already do. Only the right side compounds — and the right side needs a standard specific enough for a machine to apply.
So the question isn't whether your methodology is good. It may be.
It's whether anything but a person can use it.
We put the longer version of these questions into a short guide — the ones to answer before you talk to anybody, and the ones to ask a vendor once you do. Choosing AI for Sales: a decision aid — no form to complete.
At Axiom Sales Kinetics we've spent thirty years helping sales teams coach, learn, and sell more effectively. If you're working out where your own methodology stands, start with the sales operating system underneath it — or visit us at www.axiomsaleskinetics.com.
