# Why AI Cites Your Page but Not Your Brand Name, and How to Fix It

> AI cites your page but drops your name when it lifts a fact and retells it in its own voice. Fix it: put your brand inside the claim and earn outside mentions.

URL: https://onlyaeo.com/blog/why-ai-cites-your-page-but-not-your-brand-name · Published: 2026-10-18 · Author: Tejas Patil

- **Engines lift passages, and some rewrite them:** Claude's citations return the exact passage that supports a claim, and some engines may restate it in their own voice, so your brand can drop out when its name is not inside the sentence they take.
- **It happens often, at rates that vary by study:** Ahrefs and BuzzStream both measured cited-but-unnamed answers, and the rate differed by brand, by engine and by study.
- **The text is what buyers read:** Pew Research Center found Google users rarely click links inside AI summaries, and Ahrefs found web mentions track AI Overview visibility more closely than backlinks.
- **Fix the page and the web around it:** write the brand into the claim sentence, name your data and frameworks, set your site name and Organization markup, and earn mentions on the third-party pages engines read.
- **Measure outcomes separately:** count named, cited-only, both and neither for each engine, because each outcome needs a different fix.

## What is a ghost citation?

A ghost citation is an AI answer that links to your page as a source while its text never names your brand.

Search Engine Land calls the pattern a ghost citation.

A citation is a link to your page, shown as an inline reference, a footnote or an entry in a sources panel; a mention is your brand name in the generated text.

The gap runs the other way too. A [BuzzStream study of AI mentions and citations](https://www.buzzstream.com/blog/ai-mentions-vs-citations-study/) found only 23.1% of brand mentions were backed by a citation to the brand's own domain in the same response.

## How common are ghost citations?

Common enough that every study below found it, though the rates differ widely.

- **Ahrefs, testing its own brand.** In a [self-promotional content experiment](https://ahrefs.com/blog/self-promotional-content-ai-seo-experiment/) covering 9,886 answers from ChatGPT, Gemini, Perplexity and Copilot, Ahrefs found 43% of answers citing a page that promoted its new conference never named the conference, while only 11% of answers citing its established tool pages left out the Ahrefs name.
- **BuzzStream, across brands.** BuzzStream found 69.9% of citations to a brand's domain also named the brand; the rest were ghost citations.
- **BuzzStream, by engine.** The ChatGPT model BuzzStream tested named a brand it cited 92.7% of the time, against about 68% on Google's AI products.
- **An agency checking its own domain.** SEOquick, a Ukrainian SEO agency, counted Google AI Overview answers citing its site and found its name in 71.4% of them, while 28.6% used the page silently.

## Why do AI engines cite a page but not name the brand?

Citing and naming happen at different levels: the citation attaches to the passage that supports a claim, while the engine writes the sentence the reader sees.

### Citations attach to passages

Anthropic says Claude's citations return the exact passages that support each claim.

For plain text documents, [Anthropic's citations documentation](https://docs.anthropic.com/en/docs/build-with-claude/citations) says content is chunked into sentences that can be cited from.

Google's [guide to AI features in Search](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide) describes the same flow: ranking systems retrieve pages, then other systems review the specific information on them and show links to pages that support the response.

### Engines write in their own voice

Search Engine Land offers one possible explanation: some engines treat the linked citation as enough attribution, then present the information in their own voice.

SEOquick puts it from the engine's side: it pulls a fact off your page, drops it into the answer and attaches a link, and saying your company's name serves no purpose there.

### Long source lists leave little room for names

BuzzStream counted an average of 33 URLs per Google AI Mode response, against 3.8 for ChatGPT. Pew Research Center found 88% of Google AI summaries cited 3 or more sources, and the median summary ran 67 words.

Where your link sits in that list tracks with the odds. In SEOquick's sample, its brand was named in 98.8% of answers when its link led the source list, 96.1% at positions 2 to 3, 58.8% at 4 to 7 and 21.6% at position 8 or lower.

## What on your page makes the brand easy to drop?

A page loses its name when the cited fact reads as anyone's, or when the name is too far away or too unfamiliar for the engine to carry along.

- **Commodity passages.** Google's guidance says commodity content "is often based on common knowledge, which could originate from anyone, and typically adds little unique insight for readers". A generic tip gives the engine nobody to credit.
- **A name far from the fact.** Search Engine Land suggests a statistic buried paragraphs below the company name is probably easier to anonymize than a figure with attribution built into the sentence, and flags the idea as a hypothesis to test.
- **A brand the engine barely knows.** BuzzStream reads a citation without a mention as a sign the brand may not be discussed enough in training data. Ahrefs' gap between its new conference and its established tools fits that reading, and [building a brand entity AI engines recognize](https://onlyaeo.com/blog/build-brand-entity-ai-engines-recognize) is the longer fix.
- **A generic site name.** In Google Search, site names are generated automatically from a site's home page and references to it across the web, and Google's site name documentation warns that a generic name is unlikely to be selected unless it is an extremely well-recognized brand.

## Does it matter if AI cites you without naming you?

Yes, for commercial questions. Few readers click the links in an AI answer, so the text is where a buyer learns who you are.

Pew Research Center's [study of Google browsing data](https://www.pewresearch.org/short-reads/2025/07/22/google-users-are-less-likely-to-click-on-links-when-an-ai-summary-appears-in-the-results/) found users clicked a link inside an AI summary in just 1% of visits to pages with such a summary. When a summary appeared, they clicked a traditional search result in 8% of visits, against 15% with no summary.

Mentions also track visibility. In [Ahrefs' study of brand signals in AI Overviews](https://ahrefs.com/blog/ai-overview-brand-correlation/), web mentions correlated with AI Overview brand visibility at 0.664, against 0.218 for backlinks. Brands with the most web mentions earned up to 10 times more AI Overview mentions than the next quartile.

These are correlations and click rates: we have not seen a study comparing recall or conversion between named and unnamed citations.

## How do you measure ghost citations for your brand?

Run your buyer questions through each engine and sort every answer by whether your brand is named, cited, both or neither. A blended visibility score hides the difference.

1. **Pick questions by persona.** Each buyer persona gets a different answer, so we research personas from the brand's market and test their questions separately.
2. **Keep the engines apart.** The BuzzStream gap between ChatGPT and Google's products shows why a pooled rate misleads.
3. **Read citations and text separately.** OpenAI's [web search documentation](https://developers.openai.com/api/docs/guides/tools-web-search) says inline citations show only the most relevant references, while a separate sources field lists every URL the model consulted.
4. **Log your link position.** Given SEOquick's position data, a ghost citation at the bottom of a long list calls for different work from a ghost citation at the top.
5. **Count recommendations on their own.** A recommendation is worth more than a passing name check, so in our client reports an answer that recommends the brand is scored apart from one that just names it or just cites its page.

| Outcome | What it usually means | Fix first |
|---|---|---|
| Named and cited | Retrieval and attribution both work | Keep the cited passage current and recheck it monthly |
| Cited only (ghost citation) | Retrieval works, attribution does not | Rewrite that passage with the on-page fixes below |
| Named only | The engine knows you from training data or third-party pages, and your page is not being retrieved | Crawl access, a page that answers that exact question, clear structure |
| Neither | The engine draws on other sources for this question | The off-site fixes below, then an answer page |

For prompt sets, run counts and scoring, see [how to measure your brand's AI visibility](https://onlyaeo.com/blog/how-to-measure-your-ai-visibility-a-step-by-step-guide).

## How do you fix a ghost citation on your own page?

Make your brand part of the fact, so the sentence an engine lifts carries your name.

- **Put the brand inside the claim sentence.** Write "OnlyAEO's check of buyer questions found..." in place of "Our research found...", since Claude, for one, can cite at the sentence level (see above).
- **Do not rely on the title.** Anthropic's docs state that a document's title is passed to the model but not used toward cited content.
- **Name your data, index or framework.** Search Engine Land notes that when a study, index or methodology carries your brand's name, the finding and the brand become harder to separate. The same logic explains [why AI credits a competitor for a statistic you published first](https://onlyaeo.com/blog/why-ai-engines-credit-a-competitor-for-a-statistic-you-published-first).
- **Add what engines quote.** In the [GEO paper by Aggarwal and colleagues](https://arxiv.org/html/2311.09735v3), adding citations, quotations and statistics gave relative improvements of 30% to 40% on its Position-Adjusted Word Count metric. A branded statistic gives the engine something to quote and a name to quote with it.
- **Lead with a branded finding.** Open the page with a sentence that joins your name to your main finding.
- **Set your site name and Organization markup.** Google says WebSite structured data on the home page indicates your preferred site name. Its Organization markup guide says the markup can help Google disambiguate your organization in search results, and its sameAs property points to your profiles on other sites.

Keep markup in proportion: Google also says structured data is not required for generative AI search and there is no special schema.org markup to add.

## How do you fix it off your own site?

Get named on the third-party pages engines already cite, because that is where most citations point.

For its established brand, Ahrefs found only 6% of new mentions came from answers citing its own pages, while 94% came from third-party content. In BuzzStream's study, 80% of citations were on other people's websites.

The split is sharper on shopping questions. An analysis of ChatGPT citations for a small set of brands by The Hoth found 69.6% of citing responses pointed to the brand's own site on questions naming the brand, but only 1.3% on questions containing "best".

The Hoth states the limits itself: "It's a directional read, not a census," based on the top 30 to 50 cited pages per bucket, on ChatGPT, in the US, on a single date.

1. **Map the sites engines cite.** For clients we keep a ranked list of the outside sites AI leans on in their market that do not mention them yet, each with a suggested play: a guest post, a partnership, a listing or a review.
2. **Ask for the name next to the fact.** When a journalist or partner cites your data, ask them to write your brand in the same sentence as the figure.
3. **Keep it authentic.** Google warns that chasing inauthentic mentions around the web helps less than people expect.

See [the earned media sources AI cites](https://onlyaeo.com/blog/earned-media-aeo-sources-ai-cites) for which outlets to approach.

## When is a ghost citation fine, or even useful?

A ghost citation is fine on informational questions where your name adds nothing, and useful as proof that engines already retrieve and trust your page.

Being cited, even without your name, may still beat being retrieved and left uncited. In Ahrefs' experiment, when a conference page was retrieved but not cited, 74% of answers skipped the conference name, more than when the page was cited.

Our rule of thumb is to triage by intent:

- **Leave definitions alone.** On "what is" questions, SEOquick's point above applies.
- **Fix commercial answers first.** On "best", comparison and pricing questions, an unnamed citation hands the buyer your facts and someone else's shortlist, the pattern behind [why AI recommends your competitor instead of you](https://onlyaeo.com/blog/why-ai-recommends-your-competitor).

## How we tested this

We tested this ghost citation question the way we start every article: before drafting, we put the question to ChatGPT and Claude, read in full each page they cited, and built sections around the searches they ran.

Both ran through OpenAI's and Anthropic's APIs, with a web search tool available and each model deciding whether to use it.

- **ChatGPT answered from memory.** It did not search and cited no sources. Its advice was generic: a shorter version of the fixes above.
- **Claude ran a single search.** It cited a mix of SEO trade press, agency blogs and software vendor blogs.
- **Claude's answer was itself a ghost citation.** Its text named only the company behind the top-cited study and a data company behind another figure. The other cited publishers became "Another source says" and similar lead-ins.
- **Neither named OnlyAEO.** We had no page on this question yet.

In our latest AI visibility report, onlyaeo.com was cited in a single answer, and that answer named OnlyAEO: far too few citations for a rate. These were single API runs, not the consumer apps, which may behave differently.

## Where does OnlyAEO fit?

OnlyAEO is an AEO platform that tracks where AI names your brand and creates the content to change it.

To start, we learn the brand's voice, facts and buyers and connect publishing. Every month a fresh report shows where the brand is named, we create articles, videos and listings the client approves, and we review together what worked.

Answer pages go on the client's own domain with guardrails so they do not hurt existing search presence.

## More questions about ghost citations

### Does repeating my brand name all over the page fix ghost citations?

Unlikely. The closest evidence is the GEO paper, where Keyword Stuffing, a classic SEO tactic, did not perform well against the baseline.

### Do AI citations always support the sentence they are attached to?

No. A human evaluation of generative search engines published on arXiv found that, on average, only 51.5% of generated sentences were fully supported by citations, and only 74.5% of citations supported their associated sentence.

### Where can I see how my pages perform in Google's AI features?

Google points site owners to the Generative AI performance report in Search Console. It covers Google's own features only.

### Can a lower-ranked page still win visibility in AI answers?

In the GEO paper, the Cite Sources method raised visibility by 115.1% for websites ranked 5th in search results, while the top-ranked website's visibility fell by 30.3% on average.