When the cited page had the brand name in its title, the answer was 60.3% less likely to leave the brand out, comparing pages from the same brand.
That matters because most AI visibility dashboards count citations. A citation tells you the engine reads your page. It doesn't tell you whether the buyer reading the answer ever saw your name.
Key Takeaways (TL;DR)
42.9% of AI citations to a brand's own page didn't name the brand when the brand was missing from the page title. With the brand in the title, it fell to 12.2%.
Within the same brand, brand-in-title pages were ghosted 60.3% less often, and the gap held in 17 of 20 brands (p = 0.003).
The pattern held on all four engines. ChatGPT ghosted untitled pages most (54.2%), AI Overviews least (37.4%).
"Best X" and comparison pages had the highest ghost rate, 49.5% without the brand in the title.
Repeating the brand name on the page didn't help. Pages with 10+ mentions were ghosted 45.9% of the time.
87% of ghost answers named other brands instead. Your page did the work and a competitor got the credit.
Ghost Citations: at a Glance
Question | Answer from the data |
|---|---|
How common are ghost citations? | 43% of answers linking to a brand's site didn't name it (9,624 of 22,375, Jun to Aug sample) |
Biggest lever found | Brand name in the cited page's title: 42.9% ghosted without, 12.2% with |
Same-brand effect | 60.3% fewer ghost citations, 17 of 20 brands, p = 0.003 |
Most affected engine | ChatGPT, 54.2% ghosted without the brand in the title |
Most affected page type | "Best X" and comparison pages, 49.5% |
Weakest lever | Brand in the first ~500 characters: 43.6% vs 36.9% |
Who gets the credit | Other brands are named in 87% of ghost answers |
What Are Ghost Citations?
A ghost citation is an AI answer that cites a URL on your domain as a source but doesn't name your brand anywhere in the answer text. The link sits in the source list. The recommendation the reader acts on mentions someone else, or no one.
We're not the first to see it. Search Engine Land reported a Writesonic analysis of about 16 million brand appearances in which roughly 40% of AI citations left the source brand unnamed. That study found the gap but said the field had no clear explanation for it yet. Our goal was narrower: find something on the page that a team can change and that moves the rate.
How We Measured Ghost Citations
The data comes from the Reach platform, which runs tracked prompts for client brands every day on four AI engines. We looked at 71,275 answers that cited exactly one of a brand's own pages, across 43 brands, between January 3 and August 21, 2026. Restricting to one cited own page per answer means each answer maps to one page and one title.
Three definitions carry the whole study:
Term | How we measured it |
|---|---|
Ghost citation | The answer cites the brand's page, and Reach's answer-extraction flag says the brand isn't named. We used the extraction flag instead of text matching so brands with generic names don't skew results. |
Ghost rate | Ghost citations divided by all citations of that page type |
Brand in title | The brand name appears as a whole word in the page title, based on Reach's stored snapshot of the page |
Pooled numbers can hide a brand-mix effect, since a few large brands own most of the brand-titled pages. So we ran a second, within-brand check on every second day of the period (36,400 page citations).
Finding 1: The Brand in the Page Title Cuts Ghost Citations
When the cited page didn't have the brand in its title, 42.9% of citations were ghosts (28,264 of 65,810). When it did, the rate dropped to 12.2% (665 of 5,465). Of everything we tested, the title was the only on-page signal with a gap this size.
Cited own page | Ghost rate | Base |
|---|---|---|
No brand in title | 42.9% | 28,264 of 65,810 |
Brand in title | 12.2% | 665 of 5,465 |
A pooled gap like this can come from brand mix, so the within-brand test is the number to quote. Among the 20 brands with at least 10 citations in both groups, brand-titled pages were ghosted less often in 17. The Mantel-Haenszel stratified estimate puts the reduction at 60.3%, with a two-sided sign test of p = 0.003. Two of the three exceptions moved less than 3 points. The third had only 18 brand-titled citations.
The effect also held over time. From January to May, ghost rates were 35.8% without the brand in the title and 8.9% with it. From June to August they were 47.2% and 12.7%. Both groups rose in the summer, which may reflect a wave of new brands joining the sample in July, but the gap between them stayed.
Finding 2: The Effect Holds on Every AI Engine
Each of the four engines ghosted brand-titled pages less often. The size of the problem varied a lot by engine, which is why we report them separately.
Engine | No brand in title | Brand in title |
|---|---|---|
ChatGPT | 54.2% (5,934 of 10,953) | 18.8% (248 of 1,317) |
Perplexity | 52.3% (8,834 of 16,902) | 12.6% (112 of 891) |
Google AI Overviews | 37.4% (6,702 of 17,913) | 8.4% (132 of 1,580) |
Google AI Mode | 33.9% (6,794 of 20,042) | 10.3% (173 of 1,677) |
ChatGPT and Perplexity drop the brand name in more than half of untitled-page citations. Google's two surfaces sit lower, but even there about one cited page in three goes unnamed. If your buyers mostly ask ChatGPT, a blended score across engines will understate the problem. We saw the same split when we checked ChatGPT's Reddit citations drop against our own data: one engine's behavior can move without the others following.
Finding 3: "Best X" Pages Get Ghosted Most
Listicle-style pages (URLs with "best", "top", or "alternatives") had the highest ghost rate in the study: 49.5% without the brand in the title. Other own pages sat at 32.4%. Adding the brand to the title lowered both, to 14.9% and 11.0%.
Our read is that AI engines treat your comparison page as a trustworthy list but don't treat your own entry in it as an endorsement. The engine takes the list, names the options it trusts, and cites you as the source. This part is a hypothesis. Only 5 brands had enough brand-titled listicles for a within-brand test, and 3 of 5 showed the drop. Treat it as a strong lead to test on your own pages.
Finding 4: Repeating Your Brand Name Doesn't Help
More mentions on the page didn't mean fewer ghost citations. Pages with 10 or more brand mentions had the highest ghost rate of any group, 45.9% (22,387 of 48,732). Pages with 3 to 9 mentions sat at 28.3%, about the same as pages with none (27.8%).
Two things explain part of this. Our count includes menus and footers, so a high count often reflects site template, and the 10+ group is dominated by listicles, which get ghosted more anyway. Placement mattered a little more than volume. Pages with the brand in the first ~500 characters were ghosted 36.9% of the time, against 43.6% without, a 6.7-point gap. The title gap was 30.7 points.
Why AI Citations and Brand Mentions Drift Apart
An AI answer is assembled in two steps. The engine retrieves sources it trusts for the question, then writes an answer that names whichever brands it can tie to the claims. A page can pass the first step and fail the second. The title is one of the strongest signals for what a page is about, so a title that names the brand ties the content to the brand before any body text is read.
The cost of failing the second step is concrete. In our engine-level sample, 87% of ghost answers named other brands. The buyer reads a shortlist that includes your competitors, built partly from your page, and your name isn't on it. That's why we track AI brand mentions and LLM citations as two separate numbers for every client.
How to Reduce Ghost Citations on Your Site
Start with the pages AI engines already cite, because those are the only ones where a ghost citation can happen. Then work through them in this order:
Pull your cited pages per engine. List every URL on your domain that appears as a source in AI answers for your tracked prompts, split by ChatGPT, Perplexity, AI Mode and AI Overviews.
Calculate the ghost rate per page. Ghost rate = citations where your brand isn't named ÷ all citations of that page. Track the named count next to the rate, since the rate improves on its own when total citations fall.
Fix titles on cited "best X" and comparison pages first. They had the highest ghost rate (49.5%) and the biggest drop with the brand in the title. Add the brand in a way that reads naturally, such as "Best Fraud Detection Tools in 2026 (Compared by [Brand])".
Name the brand early in the page. The intro effect is smaller, but it costs nothing to put the brand in the opening paragraph.
Stop counting mentions. Repeating the brand name further down the page showed no benefit in our data.
Re-measure monthly. Title changes take time to reach each engine's index, so compare named citations per page before and after, engine by engine.
Titles are one signal among several. If you're also weighing structured data, we looked at whether schema markup affects AI citations in a separate analysis.
What This Study Can't Tell You
This is observational data, so it shows a strong association and doesn't prove that a title change causes the drop. Pages with the brand in the title are often homepages, product pages and "Brand vs X" pages, which are about the brand by design. We're now running a controlled test, changing titles on live client pages and tracking ghost rates before and after, to measure the effect directly.
A few other limits are worth knowing. The top brand supplies 34.5% of brand-titled citations and the top three supply 60.5%, which is why we lead with the within-brand figure. Titles come from Reach's snapshot of each page, which may differ from the version an engine saw. The sample is Reach's client base, mostly B2B and SaaS companies in Portugal and the US. Perplexity had a citation-reporting gap from mid-June to mid-July, which lowered its counts but barely changed its ratios.
Everything You Need to Know About Ghost Citations
Topic | What we found | What to do |
|---|---|---|
Definition | AI answer cites your page but doesn't name your brand | Track cited and named as two metrics |
Size of the problem | 43% of answers linking to a brand's site didn't name it | Measure your own rate per page |
Page title | 42.9% ghosted without the brand, 12.2% with | Add the brand to cited page titles |
Same-brand test | 60.3% fewer ghosts, 17 of 20 brands, p = 0.003 | Quote this figure, not the pooled one |
Engines | ChatGPT 54.2%, Perplexity 52.3%, AI Overviews 37.4%, AI Mode 33.9% (no brand in title) | Report each engine separately |
Page type | "Best X" pages 49.5%, other pages 32.4% | Fix comparison page titles first |
Mention count | 10+ mentions: 45.9% ghosted | Don't add repetitions |
Intro placement | 36.9% vs 43.6% | Name the brand in the first paragraph |
Who benefits | Other brands named in 87% of ghost answers | Check which competitors replace you |
Find Out How Often AI Ghosts Your Brand
Reach tracks every prompt your buyers ask across ChatGPT, Perplexity, Google AI Mode and AI Overviews, daily, and records citations and brand mentions separately for each page and each engine. That split is what made this study possible, and it's what shows you which of your cited pages are building a competitor's shortlist. Our team then fixes those pages for you, from titles to comparison content, and reports named citations, not just citations.
Reach is built for B2B and SaaS teams who already get some AI visibility and want it to show up as their brand name, and as pipeline. Companies like ColdIQ, Indie Campers, Carro and Fidelidade use it to turn AI search into a growth channel; see the customer results for the numbers.
Apply for a free AEO growth audit and we'll show you your ghost rate per page and engine, plus the pages to fix first. It's free, and agencies typically charge $3K+ for the same audit.
FAQs About Ghost Citations
What are ghost citations in AI search?
Ghost citations in AI search are answers that cite a page on your domain as a source but never name your brand in the answer text. In our study of 71,275 AI citations of brands' own pages, 42.9% were ghost citations when the brand wasn't in the page title. The reader sees your link in the source list but not your name in the recommendation. In 87% of these answers, the engine named other brands instead.
How do you calculate ghost citation rate?
You calculate ghost citation rate by dividing the number of AI citations of your page that don't name your brand by the total number of AI citations of that page. A page cited 100 times with your brand named in 60 of those answers has a 40% ghost rate. Calculate it per page and per engine, because ChatGPT and Google AI Mode ghosted the same type of page at 54.2% and 33.9% in our data. Track the named count next to the rate, since the rate can improve just because total citations fall.
Why does ChatGPT cite my site but not mention my brand?
ChatGPT cites your site but doesn't mention your brand when it trusts your page as a source but can't tie the content to your brand strongly enough to name it. ChatGPT had the highest ghost rate in our study, 54.2% for cited pages without the brand in the title. When the brand was in the title, ChatGPT's ghost rate dropped to 18.8%. Comparison and "best X" pages are the most affected, because ChatGPT uses them to build a list of options without counting the publisher as one.
Which AI engine has the most ghost citations?
ChatGPT had the most ghost citations in our data, with 54.2% of citations to untitled own pages leaving the brand unnamed. Perplexity followed at 52.3%, then Google AI Overviews at 37.4% and Google AI Mode at 33.9%. All four engines ghosted brand-titled pages far less, between 8.4% and 18.8%. The ranking differs between studies, so measure your own brand per engine.
What is the difference between AI citations and AI brand mentions?
The difference between AI citations and AI brand mentions is that a citation is a link to your page in the answer's sources, while a mention is your brand name written in the answer itself. A page can be cited without a mention, which is a ghost citation, or your brand can be mentioned without any of your pages being cited. Buyers act on the names in the answer, so mentions shape the shortlist. In our data, 43% of answers that linked to a brand's site didn't mention the brand.
How do you get AI to mention your brand?
To get AI to mention your brand on pages it already cites, start by adding the brand name to the titles of those pages. In our study, that change was associated with 60.3% fewer ghost citations within the same brands. Naming the brand in the first paragraph helped a little (36.9% vs 43.6% ghosted), while repeating it across the page did not. Getting named on third-party pages the engines cite is the other half of the work.
Do ghost citations still send traffic?
Ghost citations can still send traffic, because the link to your page stays in the answer's source list. They do less for your brand, since the buyer reads a recommendation that doesn't include your name and, in 87% of cases in our data, includes other brands. A click from a source link also arrives without the context of a recommendation. Measure AI referral traffic and named citations together to see both effects.
Will adding my brand to page titles hurt my Google rankings?
Adding your brand to page titles won't hurt your Google rankings if the title still leads with the keyword the page targets. Google often appends the site name in search results already, and a format like "Best Fraud Detection Tools in 2026 (Compared by [Brand])" keeps the keyword first. The risk is only a title that runs too long and gets cut off, so keep it under about 60 characters. Test on a handful of cited pages first and compare rankings and named citations after a month.
