Ghost Citations: 62 Percent of AI Citations Never Name Your Brand and That Gap Is Where You Lose

Your AI visibility report says your domain appears in 30% of relevant AI answers. You see the citation count climbing. You assume that means buyers are finding you. It does not. Semrush and Kevin Indig's 2026 study of 3,981 domain appearances across four AI engines found that 61.7% of all AI citations are ghost citations: the AI sourced your page but never said your brand name in the answer. The buyer reads the response, gets the recommendation, and walks away without knowing you were the source. Your content did the work. Your competitor got the credit.

What a Ghost Citation Actually Is

Kevin Indig coined the term to describe a specific gap. The AI engine finds your page relevant enough to include as a source link. It appends your URL in the footnotes or the source panel. But in the actual answer the buyer reads, your brand name never appears. The buyer sees three brand recommendations, clicks one, and never scrolls down to the citation list where your URL sits.

The Seer Interactive study analyzed 541,213 LLM responses across 20 brands and explained why this happens. Their finding should stop you: citations are post-hoc. The AI generates its answer first, decides which brands to recommend from its parametric memory (the knowledge baked into the model during training), and only then goes looking for source links to support those choices. The citations are the bibliography, not the brainstorm. Your page passed the retrieval threshold. Your brand did not pass the recommendation threshold. Those are two separate systems, and most brands are only measuring the first one.

The Numbers Are Not Close

The Semrush study measured citation rate versus mention rate across four engines. The results make the gap structural, not incidental.

ChatGPT cites brands in 87% of appearances but mentions them by name in only 20.7%. It operates like an academic paper: footnotes everywhere, brand names almost nowhere. Gemini does the opposite: it names brands in 83.7% of appearances but only generates a citation link 21.4% of the time. Google AI Overviews sit in the middle. Google AI Mode mentions brands at nearly twice the rate of ChatGPT but still acts closer to a footnoted research output.

Only 13.2% of all brand appearances in the study were both cited and mentioned. That is the only bucket where the buyer sees your brand name and can click through to your page. In every other scenario, either the AI says your name without linking (25.1%) or links your page without saying your name (61.7%). The citation rate across all appearances is 74.9%. The mention rate is 38.3%. Citations run at nearly double the rate of mentions. The metric most brands track is the one that matters least to the buyer reading the answer.

Engine Disagreement Means Single-Platform Tracking Is Fiction

The Semrush team tested 454 prompt-and-domain combinations across multiple engines and found that in 22% of cases, the engines disagreed on whether to name the brand. Instagram was mentioned by ChatGPT and Gemini but only cited (not named) by Google AI Overviews. Facebook was named in every Gemini appearance but named in only 1 of 9 Google AI citations.

There was almost no overlap between the brands ChatGPT cited and the brands Gemini named for the same prompts. Kevin Indig's summary: "These are different behavioral systems. Treat them that way." If you are running your AI visibility audit on one engine, you are structurally blind to what every other engine is doing with your brand.

The Brand Tax: Stranded Equity You Are Paying Right Now

Seer Interactive named the business cost precisely: stranded brand equity. Here is the sequence. A buyer asks "who are the best solutions for [your category]?" The AI reaches into its parametric memory and surfaces three competitor names. Retrieval runs. It finds a blog post from your domain that is topically relevant. Your URL gets appended as a supporting source. The response goes out: competitors recommended, your content cited, your brand silent.

You paid for the research. You paid for the writing. You paid for the editorial quality that cleared the retrieval threshold. And every dollar of that investment generated a recommendation for someone else. Seer's data showed a 5x lift in the wrong direction: when a brand IS mentioned in a response, its citation rate jumps to 53.1%. When it is NOT mentioned, citation rate drops to 10.6%. If citations caused mentions, those numbers would be similar. They are not close. The causation runs from brand authority to citation, not the other way around.

Searchless.ai's analysis of 50,000 AI responses across 500 brands found that 88% of brands were never mentioned once. The top 12% captured 94% of all mentions. The top 3% captured 51%. This is not a ranking curve. This is a cliff. You are either in the recommendation set or you do not exist.

What Actually Drives Mentions vs Citations

Mentions come from two sources: the model's training data (how often your brand appeared in the training corpus next to category terms) and third-party pages the AI retrieves (when a review site or industry report names your brand, the AI may echo that reference). AirOps documented that 58% of brand mentions in AI search results do not include a citation link. Mentions are a recognition signal. Citations are a retrieval signal. Different inputs, different systems, different optimization paths.

Searchless.ai found that brands with entity mentions across 6 or more referring domains had a 73% citation rate. Brands with mentions on fewer than 3 domains had a 6% citation rate. But the real finding is what separated cited-and-mentioned from cited-and-ghosted. The answer is not content quality. It is brand architecture: entity density across trusted domains, consistent category language, and the kind of multi-surface presence that makes the AI treat your name as a default answer.

The Semrush study added one more layer: content type matters. Comparative content ("X vs Y," "best tools for Z") produces 2.4x more brand mentions than informational content. Short, conversational queries produce 30x to 50x more brand mentions than long prompts. If your content strategy is built around long-form informational guides, you are optimizing for citations while structurally depressing your mention rate.

How to Audit Your Ghost Citation Rate

This is the part where most pieces would give you a vague framework. I will give you the concrete steps.

1. Pick your five highest-value buyer queries. Not vanity searches. The queries a buyer types before making a purchase decision.
2. Run each query across ChatGPT, Gemini, Perplexity, and Google AI Mode. Do not test on one engine. Semrush proved that engine disagreement runs at 22%.
3. For each response, record two separate columns: was your brand NAME in the answer text, and was your URL in the source list? Track these independently. The mention rate and the citation rate are different numbers.
4. Calculate your ghost citation rate: the percentage of appearances where your URL was cited but your brand name was absent from the answer. If that number is above 50%, you have a brand architecture problem, not a content problem.
5. Run the same test for your top three competitors. Compare mention rates, not citation rates. The competitor with the highest mention rate is winning the AI recommendation regardless of who has more citations.
6. Repeat monthly. Arcalea's research showed that only 2.3% of ChatGPT citations remain consistent across three identical runs. Single-snapshot audits are noise. Trends across months are signal.

The Fix Is Not More Content

More content earns more citations. It does not earn more mentions. The Searchless.ai data proves this: cited brands published an average of 12 content pieces per month. Uncited brands published 2.4. But the driver was not volume. It was entity consistency: the brand name, category terms, and competitive comparison language appeared consistently across all content and across external surfaces.

The ghost citation problem is a Machine Relations problem. It is the gap between being a source the AI retrieves and being an authority the AI recommends. I built AuthorityTech around measuring this through the Machine Relations Index, which tracks citation rates across six AI engines. But citation rates are only half the picture. The other half is whether the AI says your name when a buyer asks the category question. If it does not, your citation rate is stranded equity.

The brands that close the gap share one pattern: they build entity authority across multiple trusted surfaces before they worry about page-level optimization. PR placements, earned media mentions on industry publications, Reddit threads, review sites, comparison pages. These are not "brand awareness" in the old marketing sense. They are the training signal that tells the AI your brand is a default answer, not just a footnote.

FAQ

What is a ghost citation in AI search?

A ghost citation is when an AI engine like ChatGPT or Gemini sources your webpage as a reference link but never mentions your brand name in the answer text. Semrush's 2026 study found that 61.7% of all AI citations are ghost citations. The buyer sees the answer and the brand recommendations but never sees your name.

Why does ChatGPT cite my page but not mention my brand?

Seer Interactive's analysis of 541,213 LLM responses found that citations are post-hoc. The AI selects which brands to recommend from its parametric memory first, then retrieves supporting source links second. Your page cleared the retrieval threshold. Your brand did not clear the recommendation threshold. Those are two separate systems.

How do I measure my AI brand mention rate vs citation rate?

Run your top buyer queries across ChatGPT, Gemini, Perplexity, and Google AI Mode. For each response, track whether your brand name appeared in the answer text (mention) and whether your URL appeared in the source list (citation) as separate metrics. AirOps reports that each platform handles them differently, so single-engine testing gives an incomplete picture.

Do all AI engines treat brand mentions the same way?

No. Semrush's study found that Gemini names brands in 83.7% of appearances but cites in only 21.4%. ChatGPT reverses this: 87% citation rate, 20.7% mention rate. Google AI Mode falls between. The engines disagreed on whether to name the same brand in 22% of tested prompt-and-domain combinations.

What is the fastest way to reduce ghost citations?

Build entity authority across 6 or more referring domains with consistent category language. Searchless.ai found that brands with entity mentions on 6 or more domains had a 73% citation rate versus 6% for brands on fewer than 3. Focus on comparative content (which produces 2.4x more mentions) and earn third-party coverage that names your brand in category contexts.