AI visibility, explained

What "AI visibility" actually refers to

AI visibility is a plain idea with an awkward name. It is whether an answer engine — ChatGPT, Gemini, Perplexity, Google's AI Overviews, and the rest — names your brand when a person asks it a question your business could answer.

A search ranking tells you where your page sits on a results page a human then scans. AI visibility tells you whether your brand is inside the answer the person reads instead of scanning results at all. Those are two different surfaces, and a site can do well on one and be absent from the other.

Why it is not just SEO under a new name

Classic SEO optimises a page to rank for a query. The reader still clicks through and judges the page themselves. An answer engine does not hand the reader a list — it composes a reply, and it decides on its own which sources to draw on and name.

So the question changes. It is no longer only "does this page rank?" but "when the engine writes the answer, does it have a clear, attributable reason to mention us, and does it?" A page can rank and still never be cited, because the engine could not extract a clean, self-contained claim from it.

How you measure it

You measure AI visibility the way the engines are actually used: by asking. You take a fixed set of prompts a real buyer would type, send each to each engine, and record whether your brand is named, how it is described, and which source the engine attributes.

The method only means something if it is repeatable. A single prompt on a single day is anecdote. The same prompt set, run the same way across engines, is a measurement you can compare over time and against competitors.

• Use prompts a buyer would genuinely ask, not prompts containing your brand name — naming yourself guarantees a mention and measures nothing.
• Run the same set across each engine, so a gap is about the engine, not the question.
• Record the attribution, not just the mention: being named is one thing, being named as the source is another.

What makes a page legible to an answer engine

Answer engines reward pages they can read without guessing. That means a clear, self-contained statement of what you do and for whom; facts an engine can lift without inferring; and a machine-readable identity that says, unambiguously, who the page belongs to.

None of this guarantees a citation. No tool can promise that, and anyone who does is selling certainty that does not exist. What it does is remove the reasons an engine would have to skip you — an unreadable claim, a missing identity, a fact it cannot verify from the page.

Where to start

Start by seeing what the engines say now. Pick five questions your best customers ask before they find you, run them, and write down who gets named. That baseline is uncomfortable and useful in equal measure.

A ScanDepth audit does this methodically — a fixed prompt set across engines, scored the same way every time — and pairs it with the on-page reasons a page is or is not legible, so the result is a list of changes rather than a verdict.