The plain definition

What is AI visibility? The measured answer

AI visibility is how often AI assistants recommend a brand when asked a category question. It is measured by putting a fixed set of buyer questions to the assistants and counting who gets named, published as dated rates.

The method, briefly

How AI visibility is measured

Take the questions real buyers ask – “best mortgage broker in Sydney” – put them to the major AI assistants the same way, on a schedule, and count who gets named. The result publishes as a rate, with its date and its range.

Two measures come out, kept separate. Mention rate – how often a business is named. Citation rate – how often its site is linked as a source. A business can be named constantly and cited never; blending them hides the thing worth knowing.

According to Southlume’s July 2026 measurement, the same buying question asked four times returned 22 different businesses from ChatGPT alone – and 12 of them appeared once and never again. That is why every reading carries a band and a date. The full readings are in the Index; the method is on the methodology page.

A different measure

AI visibility vs SEO

SEO gets you ranked in the links; AI visibility gets you recommended in the answer.

SEOAI visibility
where you show upthe list of linksinside the answer
the askrank me in the resultsrecommend me in the reply
the fieldten blue linksa few names
you win byrankingbeing named

Related – good SEO often feeds it – but measured differently, and one can be strong while the other is absent. Treating AI visibility as a line item of SEO is how it goes unmeasured.

The moving target

Why it’s hard to measure well

The stability band, in one breath

Ask an AI the same question four times and the answers shift. The band is that range. A change only counts as movement once it clears it.

The method in full →
illustrative of method

AI assistants are not deterministic. Ask one the same question twice and the answers differ; ask it repeatedly and the recommendations can change substantially. A single check tells you almost nothing.

Measuring it well means repetition, fixed questions, and ranges. A reading without a date and a range isn’t a measurement – it’s a screenshot.

Loose ends

Frequently asked questions

What is AI visibility?
How often AI assistants recommend a brand when asked a category question, measured with a fixed question set and published as dated rates.
How is AI visibility measured?
By asking the assistants a fixed set of buyer questions on a schedule, counting mentions and citations separately, and publishing rates with their bands. The method is public.
How is AI visibility different from SEO?
SEO is being ranked in the links; AI visibility is being recommended in the answer. Related, measured differently.
Which AI assistants matter for visibility?
The ones your customers actually use. Southlume measures ChatGPT, Perplexity, Gemini and Claude, as at question-set version 1.
Can you track AI visibility over time?
Yes – with repeated readings using the same questions, so change is measured rather than remembered. The band separates real movement from randomness.
What is a good AI visibility score?
There’s no universal number – rates are relative and category-specific. And a zero means “not found on these questions, on this date”, never “invisible”.

Where to from here

The measured answer to “what is AI visibility” only matters if someone applies it.

see the methodology →

Definitions per question set v1, 5 Jul 2026 · market Australia.