SEO »
Around 40% of AI citations leave the source brand unnamed, exposing a blind spot in how companies measure visibility in AI search. #
Across roughly 16 million brand appearances, new data from Writesonic found that around 40% of AI citations didn’t name the source brand in the generated answer. On Perplexity, the rate reached 52%. ( Disclosure: The article co-author, Samanyou Garg, is the founder and CEO of Writesonic.)
Getting cited by an AI engine is often touted as a win. You did the work, and now your page appears in the sources, your URL shows up in citation reporting, and the engine has found your content useful enough to support its answer.
But one thing people in our field aren’t often talking about is the fact that even with a citation, the reader may never see your brand name.
We call this a ghost citation: an AI engine links to your page as a source but doesn’t mention your brand in the answer itself.
That distinction changes what a citation actually tells you. It confirms that your content contributed to the answer, but not that your brand was visible to the reader.
Key findings
- Around 40% of citations didn’t name the source brand in the answer.
- Ghost citation rates ranged from 19% to 52% across engines.
- Perplexity had the highest rate at 52%, followed by Google AI Mode at 49%.
- Google AI Overviews reached 41%, while ChatGPT reached 37%.
- Gemini, Grok, and Microsoft Copilot had lower rates at 25%, 22%, and 19%, respectively.
- The engines broadly split between those that link frequently and those that name brands more consistently.
- Citation rate and brand visibility should therefore be measured as separate outcomes.
A citation doesn’t mean your brand was seen #
A citation means an AI answer links to your page as a source. It may appear as an inline reference, a footnote, or a link inside a sources panel.
A mention means your brand name appears in the generated answer text. Consider these two answers:
With a citation and a mention, the finding is connected directly to Writesonic, and the underlying research is also linked:
- “Writesonic’s analysis of roughly 16 million brand appearances found that approximately 40% of
AI citationsdidn’t name the source brand.” But a ghost citation looks different. The Writesonic page still appears in the sources, but unless the user notices and opens the citation, there’s nothing in the answer connecting the finding to the company that produced it.
- “One analysis found that approximately 40% of AI citations didn’t name the source brand.”
Both answers could link to the same Writesonic page. But only the first one tells the reader who produced the research.
That is the ghost citation gap: your content is visible to the engine, while your brand may remain invisible to the reader.
This gap becomes measurable when citations and mentions are tracked separately. Instead of treating every source link as the same result, teams can see where they were cited, where they were named, and where the two signals separated.
[ Be the brand AI recommends. See your AI visibility
](https://www.semrush.com/ai-seo/overview?utm_campaign=ic_sel_0101ai&utm_source=searchengineland.com&utm_medium=overlay&onboarding=off) See where your brand appears in AI search, where competitors are winning, and what it takes to become the answer AI recommends.
Perplexity drops the brand name in 52% of citations #
The ghost citation rate varied sharply by engine.
Ghost citation rate by AI engine. Note: Google AI Mode (49%) is omitted for clarity. Its behavior closely mirrors AI Overviews.
Perplexity had the highest rate in the study. In 52% of its cited appearances, the engine linked to a brand’s page without naming the brand in the answer.
Google AI Mode followed at 49%, while Google AI Overviews reached 41%. ChatGPT sat slightly below the overall average at 37%.
The rate fell to 25% for Gemini, 22% for Grok, and 19% for Microsoft Copilot.
This means two brands can earn the same number of citations and receive very different levels of recognition depending on the engine.
Out of 100 cited appearances on Perplexity, the brand would be absent from the answer text in approximately 52. On Copilot, it would be absent in approximately 19.
So the citation count could look identical in a report, but the number of times the reader actually sees the brand wouldn’t.
AI engines split into namers and citers #
The difference becomes clearer when citation rate and mention rate are compared.
- Namers, including Gemini and Microsoft Copilot, are more likely to include brands in the answer text but link to their pages less frequently.
- Citers, including Perplexity and Google’s answer engines, link to pages more frequently but are less consistent about naming the brands behind them.
ChatGPT and Grok fall between the two groups. Neither outcome is ideal.
A mention without a citation makes the brand visible but may not provide the supporting source and encourage referral clicks to the website. A citation without a mention makes the page available but may leave the brand out of the response, arguably missing the more important outcome: brand recognition.
The strongest outcome is receiving both. This split also means that to improve visibility across AI engines, your optimization strategy should differ by engine, which other industry research backs up, too.
Chasing more citations on Perplexity may increase your source count without increasing how often your brand appears in the answer. On Gemini or Copilot, however, you’re more likely to be named, but you need to be cited more consistently to drive referral traffic.
Why do ghost citations happen #
Ghost citations are likely the result of how different AI engines process and synthesize sources rather than any deliberate choice to omit brand names.
But the honest answer is: our field doesn’t have a complete explanation yet.
One possible explanation is that some engines treat the linked citation as sufficient attribution and then present the information in their own “voice.”
Take this example below. Google’s AI-generated answer to “how to make GEO profitable” offers tips and takeaways similar to those in the highlighted citations, but its style, tone, and phrasing are unique.
The engine has absorbed the source materials and rewritten them, leaving the citation in the footnotes but the brand name out of the answer.
Content structure may also play a role. When a brand name isn’t closely paired with the claim it produced on the source page, synthesis has more room to sever them.
A statistic buried three paragraphs below the company name that generated it is probably easier to anonymize than one where attribution is built into the sentence itself. This is not a proven rule, but something I’d suggest testing.
What we don’t yet know is the relative weight of each factor, or whether there are others. What the engine-level variance does confirm is that the rates aren’t uniform, which gives us patterns worth investigating even if we can’t yet explain what drives them.
How to turn citations into brand recognition #
There’s no proven method that guarantees an AI engine will name a brand whenever it cites its URL. But research and data we’ve seen point to these practical steps:
- Start with measurement, because you can’t fix what you’re not tracking.
- Build proprietary data as your brand moat.
- Make attribution explicit on your pages.
- Reinforce your brand name across third-party sources.
1. Measure mentions and citations separately
Classify tracked appearances into four groups:
- Cited and mentioned.
- Cited but not mentioned.
- Mentioned but not cited.
- Neither cited nor mentioned.
Then segment those outcomes by engine. The study shows that an aggregated citation score can hide large differences between platforms.
2. Build named research your brand can own
Proprietary research or data can become your brand moat because it may be harder to ghost than generic advice. When a study, index, framework, or methodology carries your brand’s name, the data and the brand become harder to separate because the finding references you by definition.
Named research also gives journalists a stronger hook to cover, since reporters are more likely to pick up studies with a clear methodology name.
That earned coverage matters doubly now: third-party content gets cited more heavily by AI engines, so the offsite PR your named research generates compounds into more AI citations over time.
3. Pair your brand name directly with the claim, not just nearby
One likely factor worth testing is the proximity of your brand name to a key stat or finding on the page. The hypothesis is that a statistic that appears a few paragraphs away from the company name that produced it is easier to anonymize than one where attribution is built into the sentence itself.
So instead of:
- “Our analysis found that around 40% of AI citations didn’t name the source brand.”
Use:
- “Writesonic’s Ghost Citation Study found that around 40% of AI citations didn’t name the source brand.”
Named studies, datasets, indexes, and frameworks can make the relationship between an insight and its creator clearer.
That association shouldn’t live only on your own site. Independent coverage, expert commentary, customer content, and industry discussions across the sources that most influence AI answers can repeatedly connect your brand with the ideas it wants to own.
4. Investigate the answers behind the metric
The ghost citation rate identifies areas of low brand mentions, and the next step is to assess the individual answers to help diagnose what to test. For important prompts where your page is cited but your brand is missing, review:
- Which page was cited?
- What information was extracted?
- Where did the brand name appear on the source page?
- Were other brands named in the answer?
- Does the pattern repeat across related
prompts? - Does it differ by engine? Then test whether clearer attribution, consistently named research, or changes to page structure affect the result over time.
The next GEO metric is recognition #
A source link confirms that an AI engine used your content. It doesn’t confirm that the reader saw your brand. That’s why citation rate can’t be the only measure of AI visibility.
GEO measurement needs to track not only whether a company appeared somewhere in the sources, but whether it was recognized in the answer. Citation rate tells you that an engine found your content. Mention rate tells you whether your brand actually traveled with it.
We track both separately for every client because a citation your reader never sees is visibility that only exists in a dashboard.
[
If AI can’t find you, customers won’t either.
See your AI visibility
](https://www.semrush.com/ai-seo/overview?utm_campaign=ic_sel_0102ai&utm_source=searchengineland.com&utm_medium=overlay&onboarding=off) Track your visibility across AI search, uncover missed opportunities, and grow your presence where customers are asking questions.
Methodology #
Writesonic analyzed a recent 30-day window of AI answer data across thousands of brands.
The ghost citation analysis covered approximately 16 million brand appearances. The engine-level rates presented in this article cover Perplexity, Google AI Mode, Google AI Overviews, ChatGPT, Gemini, Grok, and Microsoft Copilot.
- A citation was recorded when an AI answer linked to a brand’s page as a source. A mention was recorded when the brand’s name appeared in the generated answer text.
- A ghost citation was recorded when a page was cited, but the brand wasn’t named in the answer.
Percentages were rounded. The results represent a snapshot of engine behavior during the measurement window and don’t establish the causes of ghost citations or their effect on clicks, brand recall, trust, or conversions.
Contributing authors are invited to create content for Search Engine Land and are chosen for their expertise and contribution to the search community. Our contributors work under the oversight of the editorial staff and contributions are checked for quality and relevance to our readers. Search Engine Land is owned by Semrush. Contributor was not asked to make any direct or indirect mentions of Semrush. The opinions they express are their own.