You have sat across from someone who talked like an expert and delivered nothing.
You remember the room. The deck was clean. The credentials were real. He said the words. AI visibility. Owned demand. Attribution. Full-funnel. Brand equity. He named every part of the problem with total fluency, and you nodded, because he sounded like the person who was supposed to fix it.
A year later the OTA line was still climbing. The deck was in a drawer. The words were still true and nothing had moved. And now an OTA is eating your lunch inside the AI’s answers.
You didn’t hire a fraud. That’s the part that stays with you. He wasn’t lying. He knew the vocabulary of your problem better than you did. He just couldn’t do anything about it. And you didn’t have a word for what was missing, so you blamed the budget, or the market, or yourself.
Here is the word.
Richard Feynman told a story about birds. His father would point to one and say: you can know the name of that bird in every language on earth, and when you’re done, you’ll know absolutely nothing about the bird. You’ll know something about people. What they call it. You’ll know nothing about what it does.
Knowing the name is not knowing the thing.
That is the entire distance between the two experts you have hired in your career. One of them can name the bird. The other one can explain it. When I say explain him, I mean the thing he claims to command. His method. His mechanism. Whether he can take the machine apart and show you why it runs.
Naming is recall. He can tell you owned demand beats OTA dependence because he read it somewhere and it’s correct. Explaining is mechanism. He can derive why it’s true from how demand actually forms, tell you the exact conditions where it breaks, build the infrastructure from nothing, and tell you what result would prove him wrong.
Ask the question that separates them. Not “does this matter.” He can answer that. Ask “why, and when does it fail.”
The one who names goes quiet. He has the vocabulary and nothing under it. The one who explains keeps talking, because the mechanism is right there and he built it.
Most marketing experts can be named. Almost none can be explained.
They have the deck. They have case studies, which too often are stories about the past with the mechanism removed, the outcome kept and the causal chain deleted. A real one tells you what was tried, why it worked, and what would have made it fail. Most don’t. They have certifications, which prove they can pass someone’s test and tell you nothing about whether they can build the thing the test is named after. They can say AI visibility matters. Ask them how a retrieval system chooses one property over another and watch the room go quiet. They named the bird. They cannot explain it.
And now the thing that should scare every expert who only knows the name.
A machine names better than he does.
An AI language model produces the full vocabulary of expertise, every term in perfect order, instantly, at a scale no human can match. It has read more marketing decks than any consultant alive. Whatever else it is, it is the most fluent namer of birds ever built. And fluency was the whole product. The expert whose entire offer is naming your problem in confident language has just watched a machine do that faster and cheaper than he can.
Which means that expert has already been replaced. Not next year. Now. If the whole of what he sells is the name of the problem stated well, the market for that closed the moment the machine could state it too. Recitation used to be scarce. It isn’t anymore.
The only expert left standing is the one the machine cannot replace. The one who can be explained. Who derives instead of recites. Who builds instead of presents. Who tells you in advance what would prove him wrong and then goes and tests it, because a claim you can’t falsify is just a name in a lab coat.
That is the whole of it. Name versus mechanism is now the line between a person worth hiring and a fluent imitation of that person that costs almost nothing.
So when you sit in the room again, and someone across the table is naming your problem with beautiful fluency, ask the second question. Why, and when does it fail. If the answer is mechanism, you found the rare one. If the answer is a better-dressed version of the name, you already have that at home, and it doesn’t send an invoice.
You can name him. The question you should have asked in that room, the question that would have saved you the year, is whether he can explain the thing at all.
