AI citation share is the percentage of eligible AI answers where ChatGPT, Claude or Perplexity names your own domain as a source. It's usually the smallest of the three visibility counts: in Reflexa's Index, brands named in 27-51% of answers were cited on their own domain in only 6-18% of them (Reflexa AI Visibility Index, September 2026).
Citation share is one slice of the wider question of how AI forms an impression of your brand from whatever it can find and trust. It matters on its own because being cited is the one count that proves the engine fetched your page, rather than someone else's description of you.
A citation is when the engine lists your own domain as a source for that specific answer, not just text that refers to your brand. Perplexity shows numbered citations next to the claims they support. ChatGPT with web search on and Claude with web search on show a sources panel or inline links. A mention can come from any page about you, a review site, a directory, a competitor's comparison article; a citation specifically means the engine's own sources list names your URL. The three separate counts, Named, Recommended and Cited, are explained in full in how to measure AI visibility.
Bottom line: you can be named often and cited rarely, because being mentioned needs only that someone, anyone, wrote about you where the engine looked.
Share of voice usually tracks how often you're recommended or named across a set of buyer questions; citation share tracks only whether your own domain backs the answer. A brand can hold a strong share of voice built mostly on third-party pages describing it, while its own site is cited rarely because the engine never needed to fetch it directly. The two numbers answer different questions and should be read side by side, not collapsed into one score.
Why it matters: a falling citation share next to a steady share of voice usually means the engine's preferred source about you has moved to someone else's page.
Because getting cited requires more than existing somewhere on the web, the page itself has to be fetchable, trusted and clear enough to quote on its own. A page an AI crawler can't reach, one that renders its text only in JavaScript, or one that buries the answer three paragraphs in, can still get your brand named by someone else while your own URL never makes the sources list. That gap is exactly what the Index's own numbers show.
| Count | What it checks | Reflexa Index, September 2026 |
|---|---|---|
| Named | Brand appears anywhere in the answer | 27-51% of eligible answers |
| Cited | Engine lists the brand's own domain as a source | 6-18% of those same answers |
Proof: Reflexa AI Visibility Index, September 2026 project-management edition, as reported in how to measure AI visibility.
Do this: treat the gap between your Named rate and your Cited rate as the thing to close, not the Named rate on its own.
Run the same buyer questions across engines and divide how many eligible answers cite your domain by the total number of eligible answers. An eligible answer is one where the question didn't already name your brand outright, since a question like "what are the alternatives to [you]" creates an artificial opportunity to appear.
Rule of thumb: re-run the exact same question set each time, changing the questions between checks makes the trend meaningless.
Making the page itself easy for an engine to fetch, trust and quote on its own, before anything else. That starts with access: a blocked AI crawler can't cite a page it never fetched. It continues with the page answering the likely buyer question in a self-contained first sentence, with structured data that matches the visible copy rather than contradicting it, since a mismatch gives the engine a reason to prefer a third-party page instead.
Why it matters: most of this is access and clarity, not outreach, which is why it's worth fixing before chasing reviews or directory listings.
None of this requires buying anything, and each step takes a few minutes:
Bottom line: citation share moves fastest when the fix is access and clarity on your own pages, both things you can check and change yourself.
The percentage of eligible AI answers in which an engine lists your own domain among its sources, not just where your brand gets mentioned in the text. It is tracked per engine and is usually the smallest of the AI visibility counts.
No. Named means your brand appears anywhere in the answer, and Recommended means you're one of the top picks. Cited is stricter still: the engine has to list your own URL as a source, which it does less often than it mentions a brand.
Usually because the engine is repeating what a review site, directory or competitor comparison already said about you rather than reading your own page directly. A page has to be fetchable, trusted and self-contained enough to quote before it gets listed as a source.
Confirm AI crawlers can fetch the page, write the answer to the buyer's question in the first sentence so it can be quoted on its own, match your visible copy to your structured data, and recheck with the same buyer questions a few weeks later.
The free check reads your pages the way ChatGPT, Claude and Perplexity read them — 23 checks, evidence included.