What Is AI Search Visibility?
Ranking is about being one of ten links. AI visibility is about being one of three sources named in a single answer. They are not the same problem.
For twenty years the goal was position. Get to page one, then to the top of it, and traffic followed. That model assumes a person looks at a list and chooses.
Increasingly they don't. They ask ChatGPT, Perplexity or Google's AI Overview and read one synthesised answer built from a handful of sources. There is no page two. Either you are in the answer or you are absent — and absent looks identical to not existing.
Why good SEO doesn't automatically carry over
Plenty of it does. Clear titles, real headings, fast servers and a working sitemap help both. But three assumptions behind classic SEO break down.
Rendering. Googlebot executes JavaScript. Most AI crawlers do not. A client-rendered site can rank in Google and be a blank page to ChatGPT — the single most expensive failure we see.
Access. Googlebot and GPTBot are different crawlers with different robots.txt rules. A site can be fully open to Google and entirely blocked to AI, often by accident.
Shape. Ranking rewards depth and coverage. Citation rewards extractability — short, self-contained passages that answer a question without needing three paragraphs of context around them. A 4,000-word essay can rank beautifully and offer nothing liftable.
The four questions
Everything reduces to a chain. Break one link and the rest is wasted.
1. Can it reach you?
robots.txt decides. Binary — a disallowed crawler has no route at all.
2. Can it read you?
Is the content in the HTML your server returns, or does it appear only after JavaScript runs? Also binary in effect.
3. Can it understand you?
Structured data, a declared identity, heading hierarchy, semantic HTML, a named author. This is where a model decides what your page is, and whether the source is worth trusting.
4. Can it quote you?
Are there passages that stand alone? Question-shaped headings with direct answers underneath, lists, short paragraphs that survive being lifted out of context.
The test that cuts through everything
Ask a language model what it knows about your domain. Not what it can find — what it already holds. The answer is blunt: either a reasonable summary of what you do, or some form of "I have no reliable information about this website."
That second response is common on sites with tidy markup and respectable Google rankings. It is the clearest possible evidence that traditional SEO and AI visibility are separate problems.
This is why our audit runs that check alongside the technical signals. A score tells you what to fix. The model's own answer tells you where you currently stand.
What actually moves it
In rough order of impact:
- Be reachable. Unblock the crawlers you're willing to be read by.
- Be readable. Server-render or pre-render your content.
- Be identifiable. Organization or Person schema, a real author, corroborating profile links.
- Be quotable. Answer questions in the shape of answers.
- Be current. Machine-readable dates. Undated content gets discounted.
What nobody can promise
No tool, including this one, can make a model cite you. Authority, topical fit and licensing all sit inside systems nobody outside those companies can observe. What you can do is remove every mechanical reason to be skipped — and most sites have several, usually without knowing it.
Run a free audit and see where your site stands on all four questions.