AI Citations in 2026: 62% of Domain Appearances Are Ghost References That Never Name Your Brand
A Semrush study of 3,981 domain appearances found 62% were ghost references: cited, never named. Here is what drives named citations vs invisible ones, engine by engine, with the data.
An AI citation is when a search engine like ChatGPT, Perplexity, or Google AI Mode uses your content as a source in its answer. The problem is that in Semrush's study of 3,981 domain appearances, 62% were ghost references: the page was cited and the brand name never appeared in the answer. Your content does the work. Someone else gets the credit. Or worse, nobody does.
I have spent the last year measuring how AI engines cite brands across every major platform. The data from a June 2026 Semrush study of 3,981 domain appearances confirmed what we had been seeing in our own client work at AuthorityTech: getting cited is table stakes. Getting named is the actual game. And most brands are losing it without realizing they are even playing.
What an AI Citation Actually Is (Not the Marketing Version)
An AI citation happens when an answer engine retrieves your page and presents it as a supporting source in its response. In ChatGPT, it shows up as a numbered footnote linking to your URL. In Perplexity, it appears as an inline source card. In Google AI Mode, it surfaces as a linked reference below the generated answer.
This is not a backlink. A backlink is one website linking to another. An AI citation is a machine selecting your content as evidence for its answer to a human question. The difference matters because the selection criteria are completely different. Google's ranking algorithm weighs hundreds of signals. An AI engine weighs whether your content directly answers the query with extractable, verifiable claims.
Think of it this way: a backlink is a vote. An AI citation is a witness called to the stand. The standard of evidence is higher, and the reward is proportionally larger. OpenAI counts 1.2 billion weekly ChatGPT users, and Google says AI Mode has passed 1 billion monthly active users. Behind every answer those products give is a source selection decision.
The Ghost Citation Problem: Hard Numbers From 3,981 Domain Appearances
Here is the part most AI visibility reports skip entirely.
A June 2026 study by Semrush and Kevin Indig analyzed 3,981 domain appearances across 115 prompts in 14 countries and four AI engines: ChatGPT, Google AI Overviews, Gemini, and Google AI Mode. An appearance is a domain showing up in an AI answer at all. They tracked two things separately for each one: whether the domain was cited (linked as a source) and whether it was mentioned (brand name appeared in the answer text).
The results:
- 74.9% of appearances included a citation link
- 38.3% included a brand name mention
- 61.7% were ghost citations: link present, brand name absent
- Only 13.2% produced both a citation and a mention
That 61.7% number is the one that should concern every founder reading this. It means your content can be doing exactly what it is supposed to do inside an AI engine, and your brand still gets zero recognition for it. Semrush's own comparison: the citation rate (74.9%) is nearly double the mention rate (38.3%), so any visibility report that counts citations as brand exposure overstates it. Because 74.9% of appearances carried a citation link, about four in five cited appearances were ghosts.
Medium.com was cited 16 times across the study and never once named in the answer text. Sixteen citations. Zero brand presence. That is the ghost citation problem in a single data point.
How ChatGPT, Gemini, and Google AI Mode Handle Citations Differently
Not all engines are equal, and the differences are not small.
ChatGPT operates like an academic paper. In Semrush's data, 87% of its domain appearances carried a citation link while only 20.7% named the brand in the answer text. If your content gets used by ChatGPT, you will most often get a footnote link and far less often your brand name in the actual answer that the user reads. The user sees the answer. The link sits in small type below.
Gemini does the inverse. It named the brand in 83.7% of its appearances but linked a source in only 21.4%. Gemini is conversational. It talks about brands the way a person would, weaving names into its answer. But it links a source in only about one appearance in five. You get the brand mention. You often do not get the link.
Google AI Mode falls between the two: it named the brand in 37.6% of its appearances against ChatGPT's 20.7%, nearly twice the rate, and cited in 76.3%. It operates more like a research paper with inline references. Google AI Overviews sit closest to the middle, with 84.9% of appearances cited and 61% naming the brand.
The strategic consequence is this: a single "AI citation strategy" does not exist. ChatGPT mostly uses brands as footnoted sources. Gemini mostly names brands it already knows. Google AI Mode and AI Overviews sit between the two. If you optimize for only one engine, you are leaving the others on the table.
In 100 of the 454 prompt-and-domain combinations Semrush tested (22%), the engines disagreed on whether to name the brand for the same prompt. One engine named the brand. Another used the content without naming anyone. Kevin Indig's reading of the study: any AI visibility report that aggregates across LLMs is misleading. You cannot assume visibility on one platform translates to visibility on another.
Why Traditional SEO Metrics Miss the Citation Signal
This is where the old playbook breaks.
An Ahrefs study of 75,000 brands measured how traditional marketing signals correlate with brand mentions in Google AI Overviews. The findings inverted decades of SEO orthodoxy:
| Signal | Spearman correlation with brand mentions in AI Overviews |
|---|---|
| Branded web mentions | 0.664 |
| Branded anchors | 0.527 |
| Branded search volume | 0.392 |
| Backlinks | 0.218 |
Backlinks, the signal that powered SEO for 20 years, correlate at 0.218 with AI Overview brand mentions, about a third of branded web mentions' 0.664. In Ahrefs' later 75,000-brand study across ChatGPT, AI Mode and AI Overviews, YouTube mentions showed the strongest correlation with AI visibility, at about 0.737. Correlation is not causation, but the direction is consistent.
This is not a minor adjustment. This is a structural inversion. The signals that made brands visible in Google's ten blue links are not the signals that make brands visible in AI answers. If your strategy is still centered on building backlinks, you are optimizing for a system that barely registers in the one where your buyers are increasingly looking.
Five Signals That Drive Named AI Citations Over Ghost References
Seer Interactive analyzed 541,213 LLM responses across 20 brands and six AI platforms and describes ghost citations as the gap between clearing the retrieval threshold and clearing the mention threshold. The five signals below are our working synthesis of what changes whether a model names a brand, not a ranked list from Seer's data.
1. Brand name is grammatically central to the claim, not an afterthought.
Seer's leading hypothesis is that LLMs decide which brands to name from parametric knowledge first, then retrieve supporting sources afterward (Seer calls this strongly supported behavioral evidence, not proven architecture). If your brand name is not already embedded in the training data as the entity that owns a specific claim, the model will use your content without naming you. The fix: stop writing "five approaches to compliance" and start writing "[Brand]'s approach to compliance," so the brand becomes inseparable from the idea.
2. The entity graph is complete and consistent.
Wikidata entries, Wikipedia presence, Organization schema, and Author schema all feed the knowledge graph that LLMs use to resolve brand identity. Inconsistent naming across these surfaces creates ambiguity. Ambiguity produces ghost citations because the model is not confident enough to attach a name to the claim.
3. Third-party sources mention your brand in recommendation contexts.
Analyst reports, earned media coverage, and industry roundups that use your canonical brand name in the context of recommending you are direct training signals for future model versions. This is the lever we expect to matter most for converting ghost citations to named citations. Search Engine Land reported this shift: PR is becoming a training data strategy, not just a media relations strategy. Spin Sucks documented the same conclusion independently, calling earned media "the keys to visibility" in the AI search era.
4. Content structure makes claims extractable.
AI engines do not read pages the way humans do. They extract discrete claims from structured content: clear H2 sections, comparison tables, direct answer paragraphs, FAQ blocks with specific numbers. If your valuable insight is buried in the middle of a 2,000-word narrative without structural markers, the engine may cite you but will synthesize the claim in its own words rather than attributing it to you by name.
5. Cross-domain corroboration creates citation confidence.
When the same claim appears across multiple authoritative domains with consistent attribution, the model treats it as established fact attached to a specific entity. A single blog post making a claim is one signal. That same claim cited in Forbes, repeated in TechCrunch, and verified in your own research creates a corroboration chain that LLMs follow when deciding whether to name the source.
How to Audit Your Brand's AI Citation Presence Right Now
Stop reading and do this.
Open ChatGPT, Perplexity, Google AI Mode, and Claude. Search the queries that matter most to your pipeline. Not your brand name. The problems your buyers type when they do not know you exist yet.
For each query, track two things:
- Citation: Does your URL appear as a source link?
- Mention: Does your brand name appear in the answer text?
A citation without a mention is a ghost citation. Your content did the work. Your brand got nothing. Count them. The ratio between citations and mentions is your ghost citation rate.
Seer Interactive recommends tracking this as a "Competitive Ghost Citation Rate", measured monthly, segmented by platform and funnel stage. In their data, the awareness stage showed a 5.0% competitive ghost rate (a narrower measure than Semrush's: the brand's URL is cited, the brand is not named, and a competitor is named in the same response), which Seer calls the most damaging place for it to happen. That is where buyers form their first impression of which brands matter.
Semrush's sample ran at 61.7% of appearances. If your ghost citation rate is that high or higher, your content strategy is working and your brand strategy is not. The content is earning citations. The brand is not earning recognition.
The Earned Media Advantage in AI Citations
Here is the connection that most citation guides miss entirely.
The Ahrefs data showing branded web mentions (0.664 with AI Overview brand mentions) and YouTube mentions (about 0.737 across ChatGPT, AI Mode and AI Overviews) as the strongest correlates of AI visibility points to one conclusion: the brands that get named in AI answers are the ones with the most third-party corroboration across the web.
This is not something you can build with on-site content alone. You cannot mention your own brand on someone else's domain. You need earned media. You need press coverage in publications that AI engines already trust. You need analyst mentions. You need the corroboration chain.
In Seer Interactive's data, a brand's citation rate was 53.1% in responses that mentioned it and 10.6% in responses that did not. That is a 5x difference. Seer reads it as mentions coming first and citations following: brands the model already names get cited about five times as often as the ones it has to look up. And the way AI learns your brand is through repeated, attributed mentions across authoritative third-party sources.
The timeline is not instant. In one client case Seer documented, brand mentions on an updated page were still at zero 29 days after the change, and Seer's model projected full effect around week eight, because models update gradually rather than re-ranking in real time. That is one case and one projection, not a benchmark.
This is why earned media is a compounding asset in the AI era. Every placement that mentions your brand by name in the context of a specific capability becomes a training signal. Those signals accumulate across model versions. The brands that started building this two years ago have a structural lead that gets wider every quarter.
What Machine Relations Changes About the Citation Equation
The data in this piece points to something that traditional PR and traditional SEO both miss.
PR thinks about placements. SEO thinks about rankings. Neither framework accounts for the fact that AI engines are now the primary reader for most business content, and those engines make independent decisions about whether to name your brand based on entirely different signals than either discipline was built to optimize.
This is the problem that Machine Relations exists to solve. It is the discipline of systematically earning named AI citations through entity authority, earned media corroboration, and structured extractability. Not optimizing for one engine. Not optimizing for one signal. Building the full evidence chain that makes the machine confident enough to put your name on the claim.
The 62% ghost-citation share in Semrush's sample is not a content problem. It is a recognition problem. Your content is already good enough to get cited. The question is whether the machine knows your brand well enough to name you.
That is a fundamentally different problem than "rank higher" or "get more coverage." It requires a fundamentally different discipline.
FAQ
What is a ghost citation in AI search?
A ghost citation occurs when an AI engine like ChatGPT or Perplexity links to your page as a source but never mentions your brand name in the answer text. A June 2026 Semrush study found that 61.7% of 3,981 domain appearances were ghost citations. The user sees the answer without knowing your brand produced the underlying content. The measured unit is the share of all logged domain appearances that were cited as a source but never named; because 74.9% of appearances carried a citation link, that is about four in five of the cited appearances. This is attribution-behaviour evidence in an observed sample; it does not establish a universal ghost-citation rate, a provider's naming rule, citation causation, recommendation lift, pipeline, revenue, or any other business outcome.
How do I track AI citations for my brand?
Search your most important buyer queries across ChatGPT, Perplexity, Google AI Mode, and Claude. For each result, record whether your URL appears as a source (citation) and whether your brand name appears in the answer text (mention). Track the ratio monthly. HubSpot offers free tracking templates for monitoring citations across multiple AI engines using GA4.
Which AI search engine gives the most brand name citations?
Gemini named the brand in 83.7% of its appearances but linked a source in only 21.4%. ChatGPT is the opposite: a source link in 87% of its appearances but a brand name in only 20.7%. If you want brand name visibility, Gemini is strongest. If you want source links, ChatGPT delivers them more consistently. Google AI Overviews (84.9% cited, 61% named) and AI Mode (76.3% cited, 37.6% named) sit in between.
Do backlinks still help with AI citations?
In Ahrefs' 75,000-brand study, traditional backlinks correlated at 0.218 with brand mentions in Google AI Overviews; branded web mentions correlated at 0.664, about three times as strongly. Backlinks still matter for traditional Google rankings, but in this correlation data they are not the signal most associated with AI Overview brand mentions.
How long does it take to improve AI citation rates?
In one client case Seer Interactive documented, brand mentions on an updated page were still at zero 29 days after the change, and Seer's model projected full effect around week eight, because models update gradually rather than re-ranking in real time. That is a single case and projection, not a benchmark.
What is the difference between an AI citation and an AI brand mention?
A citation is a source link to your URL. A mention is your brand name appearing in the answer text. Only 13.2% of AI appearances produce both. A citation without a mention means your content is being used but your brand receives no recognition in the answer the user actually reads.