Some of the best consultants, strategists, coaches, and specialists in their categories are invisible to ChatGPT, Perplexity, Google AI Overviews, and Grok. Not because they’ve done anything wrong. Because the depth and quality of their work lives in private: in the precision of their diagnostics, in the judgment calls no one else saw them make, in client relationships that never became public testimonials. The machine has nothing to read.
And in their category, someone with a tidier LinkedIn, a clearer website, and a few posts that used the right language is getting cited.
This is how AI recommendation works. You can be outraged by it, or you can understand it. Understanding it is what makes it fixable.
I watched this exact gap close for a client. Andre Obradovic runs a beautiful, expensively built men’s health coaching site, and the engines couldn’t see him at all. The fix wasn’t a redesign or a pile of generic content. Both would have buried him under the same beige description every coach gets. It was getting him named as himself: a credentialed ICF and Primal Health coach with his actual positioning, in a form the machines could verify. After about six weeks of that work, all four engines (ChatGPT, Perplexity, Google AI and Grok) name him first in most when potential clients ask who to hire. His page views are up 51% and his email leads doubled month on month (GA4, May to June). For a coach who offers a premium program, being the first name an engine hands a client ready to hire isn’t a traffic stat. It’s a client he’d never otherwise have met. Same coach, same quality of work. The only thing that changed was what the machines could read. The full case study is here.
What does legibility mean in AI search?
Legibility means an AI engine can read your public material and form a confident, specific, accurate picture of you: what you do, who you do it for, and what makes you the right choice for a particular problem. Engines like ChatGPT, Perplexity, Google AI Overviews and Grok can’t access your client outcomes, your private proposals, or the quality of your judgment. They work entirely from public, crawlable, quotable material. So they don’t recommend the best business in a category. They recommend the one they understand most clearly. A less accomplished business the engine can read precisely will consistently outrank a better one whose signals are vague, scattered, or buried. Legibility is what separates the two, and the gap is fixable.
It’s a precise word for a simple idea: how well an AI engine understands you. I use it throughout because it’s the technical term, but every time you read it, you can read “how clearly the engine understands me.”
It’s not the same as visibility. You can be visible (showing up in searches, appearing in directories, having a LinkedIn following) and still be illegible to an AI recommendation engine.
Legibility is whether the engine can form a confident, specific, accurate picture of you from the public material it has access to. Not a vague impression. A specific, repeatable description. Something like: “This person helps founder-led service businesses get found by AI engines. They do this through a structured process of improving legibility, consistency, and quotability of public signals. They work with established service businesses, not startups.”
If the engine can construct something like that, with high confidence, from your public pages, profiles, and content, you’re legible. If it gets something vague and general, or contradictory, or incomplete, you’re not.
Most founder-led service businesses fall somewhere in the middle. They’re partially legible: the engine can tell roughly what category they’re in, but not specifically enough to cite them over someone clearer.
What are the two most common legibility failures?
Vague positioning
The version that shows up most often is positioning that’s aspirational rather than descriptive. “I help businesses thrive.” “Strategic partner for growth.” “Helping you reach your potential.”
These phrases mean something in a relationship, when someone has already chosen to work with you and you’re setting a tone. They mean nothing to a machine trying to decide whether you’re the right person for a potential client who asked “who helps with [specific problem].”
The engine can’t connect “helping you reach your potential” to any specific problem. It can’t form a precise picture of who you work with or what you actually do. So it doesn’t cite you. It cites the person who wrote “I help founder-led consultancies improve their AI search visibility by fixing the structural gaps that stop them from being cited.”
That second version is less lyrical. It’s also cited, and the first one isn’t.
This isn’t a call to write dull copy. It’s a call to have at least one layer of your content that’s specific enough for a machine to parse. You can have a lyrical headline and a specific service description. The engine will read the specific one.
Scattered or contradictory signals
The second failure is when different sources describe you differently.
Your website positions you one way. Your LinkedIn reflects a role you held two years ago. An industry directory lists the category you used to work in. A podcast episode that ranks for your name describes a business you’ve since pivoted. An old guest post talks about services you no longer offer.
The engine gathers all of these and tries to form a unified picture. When the signals are pulling in different directions, the picture blurs. A blurred picture is a less confident recommendation. And a less confident recommendation often gets dropped in favour of someone clearer.
The people who feel this most acutely are founders who’ve repositioned. The old signals persist for longer than you expect. Google’s index, other sites linking to old content, directories that don’t auto-update. The machine keeps reading the old you alongside the new you and averaging them together, which produces a picture that’s accurate to neither.
This is the layer where Andre was failing. He holds genuine ICF and Primal Health certifications, but his site never stated them anywhere a machine could check, and the page meant to anchor his location was simply headed “Location.” Tying his name and his certifications back to the bodies that issue them, then rebuilding his Google Business Profile so every source told one story, was what let ChatGPT, Perplexity, Google AI and Grok confirm he was a real, credentialed person and start naming him.
Entities, not keywords covers why the accumulated picture matters and how the engines build it.
How do you make your business legible to AI?
It’s three things, in roughly this order.
Name what you do specifically. Not in your tagline, if that’s not where you want to be specific. But somewhere visible, early in your key pages, in language that matches what a potential client would search for. The aim is for the engine to be able to read your homepage and complete the sentence “this business helps _____ with _____ by _____” without interpreting or inferring anything.
Make your signals agree. Audit your public presence. Every profile, bio, directory listing, and page where you’re described. Pick the sentence that’s true about you now and make sure everything public says a version of it. This is maintenance work and it matters disproportionately.
Structure your content for quoting. Clear headings, answers near the top of sections, specific claims rather than general ones, and genuine FAQs that answer what clients actually ask. The engine needs somewhere to lift the reason it’s going to attach to your name.
None of this requires you to water down your expertise or become someone you’re not. The opposite, in fact. The best legibility work makes you more precisely yourself, not less. It takes the expertise that’s been living in private and gives it a structured public form.
This is where legibility and identity stop being separate things. A business that’s clearly legible to an AI engine is almost always clearer to human readers too. The same work that makes you citable makes you easier to refer.
For founders who want to get cited as themselves, not as a beige composite of their industry category, the next piece is Get cited as yourself. The legibility work and the voice work are the same job, and that’s where the two connect.
If you want to see where your specific gaps are and what to fix first, the $197 Citeability Snapshot is the place to start, and its fee credits toward the AI Visibility Audit if you go on to the full work. It shows you the exact answer the engines give right now, a baseline score, and a ranked set of fixes you can start the same day.