ObsurfableGetting your URL cited in an AI engine response is not the same as getting your brand mentioned. And the gap between those two outcomes is larger than most marketing teams realize.
In a study published in June 2026, Semrush in partnership with growth advisor Kevin Indig analyzed 3,981 domain appearances across ChatGPT, Gemini, Google AI Overviews, and Google AI Mode, using 115 prompts across 14 countries. The core finding: 61.7% of those appearances were what Indig calls ghost citations — the source link surfaced in the AI response, but the brand name was never spoken or written anywhere in the body text.
If you have been measuring AI visibility by counting citations alone, you have been measuring the wrong thing.
What is a ghost citation?
A ghost citation occurs when an AI platform includes a URL as a source but never mentions the brand name in the response text. The domain gets traffic attribution. The company gets zero name recognition from the people reading the answer.
The reader sees the claim and maybe a link icon. They never see who said it. Your content powered the answer. Someone else — or no one — got the credit.
Kevin Indig coined the term in April 2026 through Growth Memo, and the Semrush study was the first to quantify it at scale across multiple engines.
The numbers: citation rate vs mention rate
The 3,981 appearances split into three categories:
| Outcome | Share of appearances |
|---|---|
| Ghost citation (cited, not named) | 61.7% |
| Both cited and mentioned | 13.2% |
| Mentioned without citation | 25.1% |
Adding these up tells the story:
- 74.9% of all appearances included a citation (ghost + named-and-cited)
- 38.3% of all appearances included a brand mention (named-and-cited + named-only)
The citation rate is roughly double the mention rate. A link is about twice as likely as a name. Of every 100 times an AI engine links to a domain, roughly 62 of those link appearances happen without the brand name appearing in the response.
The rarest outcome — the one most teams assume is the default — is both citation and mention together, at just 13.2%.
Each engine behaves differently
The aggregate numbers hide sharp platform-level differences. ChatGPT and Gemini are effectively opposites:
| Platform | Citation rate | Brand mention rate | Pattern |
|---|---|---|---|
| ChatGPT | 87.0% | 20.7% | Citation-heavy, ghost-prone |
| Gemini | 21.4% | 83.7% | Mention-heavy, low citation |
| Google AI Overviews | Mid-high | Lower | Significant ghost gap |
| Google AI Mode | Mid range | ~2× ChatGPT (~41%) | Moderate ghost gap |
ChatGPT cites aggressively but rarely names brands — a +66.3 percentage point gap between citation and mention rates. Your page is likely to appear as a source link; your brand is unlikely to appear in the answer text.
Gemini does the reverse — it names brands 83.7% of the time but cites them only 21.4% of the time. A −62.3 percentage point gap in the opposite direction. Users see your name but may not get a clickable source.
Visibility in one engine cannot be assumed to carry over to another. A brand dominating ChatGPT citations may be invisible in Gemini mentions, and vice versa. This aligns with CiteMetrix's finding of only 11% domain overlap between ChatGPT and Perplexity — see Separate AEO Strategies for ChatGPT and Perplexity.
Query type determines ghost rate
Not all queries produce ghost citations at the same rate. Intent shapes whether the AI names brands or only links sources:
| Query type | Citation rate | Mention rate | Notes |
|---|---|---|---|
| Informational ("what is," "how does") | 89.3% | 18.0% | Highest ghost rate |
| How-to | Moderate | 42.8% | Better mention rate |
| Comparative ("best," "vs," "alternatives") | Lower | 43.3% | 2.4× more mentions than informational |
| Commercial ("pricing," "buy," "cost") | 84.4% | 35.6% | High citation, moderate mention |
Informational queries are the ghost citation trap. AI engines love citing authoritative sources for definitional content — but they summarize the information without naming the source brand. An "what is AEO" query may cite your comprehensive guide while attributing the definition to no one.
Comparative and evaluation queries produce dramatically more brand mentions. "Best CRM for startups" or "HubSpot vs Salesforce" forces the AI to name specific products as part of the answer structure. Indig's study found evaluation and comparison content generates 30× more brand name mentions than purely informational content.
This has direct content strategy implications. If your AEO program is built entirely on informational guides, you are optimizing for the query type most likely to produce ghost citations.
Why ghost citations happen
Three mechanisms drive the gap:
1. Citation is a retrieval decision; mention is an entity decision.
When an AI retrieves a passage, it may use the information while treating the source as background evidence. Naming the brand requires a separate entity-resolution step — recognizing that a specific company produced the insight and deciding it is relevant to name in the answer. That step does not always fire.
2. Informational content structurally favors ghosts.
Definitional and explanatory content gives the AI facts to synthesize. The AI's job is to answer the question, not to credit every source. Unless the content explicitly positions the brand as the authority ("At Obsurfable, we define AEO as..."), the model has no reason to name it.
3. Platform architecture shapes defaults.
ChatGPT's architecture prioritizes source transparency (high citation rate) but does not require brand attribution in answer text. Gemini's architecture leans toward entity-rich answers (high mention rate) but does not always link to sources. Each platform's design choices create different ghost profiles.
What ghost citations cost you
Brand recognition. Users who do not click through received no brand signal from the interaction. Your content did the work; your competitor's name may appear in the answer text sourced from your data.
Misallocated budget. Teams celebrating citation count growth may be overestimating real visibility. A 20% increase in citations with flat mention rate is a 20% increase in ghost citations — more work for the same brand exposure.
Competitive vulnerability. If your comparison page powers an AI answer that names your competitor but only links to you as a footnote, the competitor wins the recommendation frame. The user remembers who was named, not who was linked.
Wrong optimization targets. Optimizing for citation rate alone pushes teams toward informational content — the highest-ghost category. Mention rate optimization requires a different content portfolio.
What to measure: four separate KPIs
AEO programs should treat citation and mention as independent axes:
| KPI | Definition | What it tells you |
|---|---|---|
| Citation rate | URL appears in source links | Content is retrievable and trusted as evidence |
| Mention rate | Brand name appears in answer text | Users see your brand in the response |
| Mention + citation overlap | Both in the same answer | Full visibility — named and sourced |
| Recommendation position | Where your brand appears in ranked lists | Competitive framing quality |
Track all four on the same prompt set, by platform, over time. Collapsing them into one "AI visibility" number makes ghost citations invisible.
The right primary metric depends on your business:
- Publishers and research sites: Citation rate is your baseline. Source links naturally outnumber name mentions for this content type.
- B2B SaaS and consumer brands: Mention rate is your priority. Users build recognition from seeing the name in the answer, not from a link icon they may never click.
How to reduce ghost-citation risk
1. Increase explicit brand entity clarity in key passages.
Do not bury brand attribution in footnotes. Lead comparison sections with "According to [Brand]..." or structure content so the brand name is inseparable from the insight. Place direct answers with brand context in the first 30% of the page — see The First 30% of the Page.
2. Shift content mix toward comparative and evaluation formats.
Build "best of" lists, head-to-head comparisons, pricing breakdowns, and alternative guides. These query types produce 2.4× more brand mentions than informational content and 30× more than pure definitional pages.
3. Strengthen third-party brand-name corroboration.
AI engines name brands that appear consistently across multiple trusted sources. Pursue G2 reviews, industry roundups, earned media placements, and Reddit discussions where your brand is explicitly named — not just linked.
4. Audit answers for wording, not just source cards.
When monitoring AI responses, read the answer text. Ask: "Is my brand named? Is a competitor named instead? What is the recommendation framing?" Source card presence alone is insufficient.
5. Optimize by platform.
On ChatGPT (87% citation, 21% mention): prioritize getting named in answer text through comparative content and entity signals. On Gemini (84% mention, 21% citation): prioritize being the brand the AI recommends, even without a source link.
Connection to the broader evidence base
Ghost citations are not an isolated finding. They sit inside a larger pattern of measurement gaps in AI visibility:
- The July 2026 GEO evidence review found no stable cross-platform optimization effects — citation and mention behave differently across engines for the same reason tactics do not transfer.
- SE Ranking found only 11.53% overlap between AI Mode advertisers and cited sources — paid visibility does not produce mention or citation lift.
- Citation sets rotate weekly (56% churn in AI Mode, 74% in ChatGPT) — a ghost citation on Monday may be a full mention on Thursday, or disappear entirely.
The through-line: AI visibility is multi-dimensional. Single-metric dashboards create false confidence.
How Obsurfable fits
Obsurfable tracks brand mentions and cited sources as separate signals across your defined prompt sets. When your citation rate climbs but mention rate stays flat, you see the ghost citation gap directly — not as a surprise in a quarterly review.
The Visibility Director can flag prompts where competitors are named in answers and you are only cited (or invisible entirely), then draft comparative content designed to close the mention gap.
For measurement fundamentals, see How to Measure AI Visibility: KPIs Beyond CTR.
FAQ
Is a ghost citation worthless?
Not entirely. The source link still drives some click-through traffic, and the citation signals that your content is retrievable and trusted. But for brand building and competitive positioning, mention rate matters more. A ghost citation is half a win.
Can I have a brand mention without a citation?
Yes — 25.1% of appearances in the study included a brand mention without a citation link. This is common on Gemini (83.7% mention rate, 21.4% citation rate). Users see your name but cannot click through to your site.
Should I stop writing informational content?
No. Informational content builds topical authority and feeds retrieval. But balance it with comparative and evaluation content that forces brand naming. A portfolio heavy on "what is" guides will produce high ghost rates by design.
How do ghost citations relate to zero-click search?
They compound it. In zero-click scenarios, users never visit your site at all. In ghost citation scenarios, users see an AI answer influenced by your content but never see your brand name. Both reduce brand exposure — ghost citations are zero-click's quieter cousin.
Bottom line
If users never see your name, your source did the work and someone else got the credit. 62% of AI citations are ghost citations. Track citation rate and mention rate as separate KPIs, optimize content mix toward comparative queries, and read answer text — not just source cards. Citation without mention is not visibility. It is attribution without recognition.