AI visibility

Why a competitor appears and you do not

Almost always because the evidence about them agrees with itself and the evidence about you does not.

An assistant builds an answer from sources that corroborate each other. Your own website is one claim — and the one it trusts least, because it is you describing yourself. Weight comes from independent agreement: a directory with the same address, a review platform with the same phone number, a trade publication that quoted you.

When those agree, the model has a fact. When they conflict — an old suite number here, a name that shortened three years ago but not everywhere — it has ambiguity, and it resolves ambiguity by naming a business it is surer about.

How this differs from ranking on Google

A results page had ten positions and a second page. Ranking eighth was not good, but eighth was visible, and plenty of businesses were built on eighth.

An AI answer has no eighth position, because the list is not long enough to have one.

So the question stops being where do I rank and becomes binary. And the omission is silent: nothing tells you that you were left out, and nothing tells the customer the list was partial. The full argument is here.

LLMO, GEO, AEO — the same job

LLMO (Large Language Model Optimization), GEO (Generative Engine Optimization) and AEO (Answer Engine Optimization) describe the same work under three names. AI SEO usually means traditional SEO with AI tooling, which is a different thing.

Anyone insisting on a hard distinction between the first three is generally defending terminology rather than explaining a method. If you want the detail: what LLMO is, and how it compares with SEO — including the 80% that carries straight over, which is the part most vendors will not tell you.

What decides whether a model names you

Entity clarity. Whether every system agrees on who you are, what you do and where. One business described one way, everywhere.

Corroboration. Independent sources saying the same thing. This is the heaviest factor and the one least under your direct control, which is exactly why it carries weight.

Structure. Facts stated so a machine reads them without inferring — the difference between a model guessing what you do and knowing it.

Answerability. Content that answers real questions in the first two sentences. Models quote what is quotable. The practical version of this is here.

How it is measured

Not by rankings. By presence: ask the assistants what your customers ask, record whether your name comes back, and record which sources produced the answer.

Measure each assistant separately. ChatGPT, Gemini, Claude and Perplexity do not share an index and routinely disagree — being known to one is not being known. And measure the same prompts repeatedly, because a single result is a snapshot and answers vary between runs.

How long it takes

Consistency fixes — name, address, structured data — can register within weeks, because they only need re-crawling.

Corroboration is slower. Getting listed, reviewed and cited by independent sources is months of work, and there is no way to buy the time back. Anyone promising AI visibility in thirty days is describing the fast half and omitting the half that matters.

Where to go next

What is LLMO — the definition and mechanics.
SEO vs LLMO — what carries over and what is genuinely new.
How to get cited by AI — the practical moves.
Working with a consultant — what the engagement looks like.
How to choose one — eight criteria, including when to hire nobody.
Current news — what is changing, updated daily.

Common questions

What is AI visibility?
AI visibility is whether an AI assistant names your business when someone asks it for a business like yours. Unlike search ranking there is no position to hold — a typical answer contains three or four names, so you are either in it or absent. It is determined by evidence about your business across sources you do not own, not by your website alone.

Why does ChatGPT recommend my competitor and not me?
Usually because the evidence about them is more consistent and better corroborated, not because their business is better. Assistants build answers from sources that agree with each other. If your name, address or phone number differs across listings, or you are absent from the platforms the assistant reads, it resolves that ambiguity by naming a business it is more confident about.

How is AI visibility different from Google ranking?
A Google results page has ten positions and a second page, so ranking eighth is still visible. An AI answer names three or four businesses and has no second page, which makes visibility binary. The omission is also silent: nothing tells you that you were left out, and the customer is not told the list was partial.

How is AI visibility measured?
By presence rather than position. Ask the assistants the questions your customers ask, record whether your business is named, and record which sources the assistant cited. Measure each assistant separately, because ChatGPT, Gemini, Claude and Perplexity do not share an index and frequently disagree, and repeat the same prompts over time because answers vary between runs.

How long does it take to improve AI visibility?
Consistency work — name, address, phone and structured data — can register within weeks, since it only requires re-crawling. Corroboration is slower: getting listed, reviewed and cited by independent sources takes months and cannot be accelerated by spending. Anyone promising results in thirty days is describing the fast half and omitting the half that matters.

Can I pay to appear in AI recommendations?
You can buy an advertisement; you cannot buy the recommendation. ChatGPT now carries ads sold through a self-serve platform, and OpenAI states they appear as clearly labelled units below the response without influencing the answer. Being named in the answer itself is still decided by evidence, not budget.