- Sites with more than 32,000 referring domains are 3.5x more likely to be cited by ChatGPT than low-authority domains, according to SE Ranking’s study of 129,000 websites and 216,524 pages.
- Nearly 50% of ChatGPT citations go to pages that rank number one on Google, with citation rates 3.5x higher for top-20 Google results compared to pages beyond position 20.
Almost 50% of ChatGPT citations came from pages ranking #1 on Google.
The citation rate is 3.5 times higher than for pages ranking beyond Google’s top 20 results.
What does this mean for your website?
Let’s break it down:
AirOps recently published data that explains why so… https://t.co/Q0t3XxPgYs pic.twitter.com/M4JZ22OdD9
— Alex Groberman (@alexgroberman) June 18, 2026
Nearly half of all ChatGPT citations come from pages that already rank number one on Google.
That is one of the headline findings from a new analysis of 129,000 domains, 216,524 pages and 100,000 prompts conducted by SE Ranking.
According to research from SE Ranking, pages ranking inside Google’s top 20 results were cited 3.5 times more often than pages outside the top 20.
The findings suggest that AI visibility may be far more dependent on traditional SEO signals than many marketers expected.
ChatGPT Still Favors Google’s Winners
The study examined 129,000 domains, 216,524 pages and 100,000 prompts across 20 different industries.
One pattern appeared repeatedly throughout the data: websites that already perform well in Google’s ecosystem are significantly more likely to become citation sources in ChatGPT.
Nearly 50% of citations came from pages holding Google’s number-one ranking.
That single finding challenges one of the more common assumptions surrounding AI search optimization. While AI interfaces may look different from traditional search engines, the sources powering those answers appear heavily influenced by many of the same authority signals that drive search rankings.
For publishers and marketers, the implication is straightforward. If a page struggles to rank in Google, its chances of being cited in ChatGPT appear significantly lower.
Backlinks Remain the Strongest Signal
If there was one clear winner in the study, it was backlinks.
Referring domains emerged as the strongest predictor of whether a website would be cited by ChatGPT.
According to Search Engine Journal’s coverage of the findings, sites with fewer than 2,500 referring domains averaged between 1.6 and 1.8 citations. Sites with more than 32,000 referring domains averaged 5.6 citations, while domains exceeding 350,000 referring domains averaged 8.4 citations.
That 32,000-domain threshold appears particularly important. Below it, citation rates clustered relatively closely. Above it, citation visibility increased sharply.
Domains with strong trust scores, largely influenced by their backlink profiles, received nearly four times as many citations as low-trust competitors.
The findings directly challenge the idea that AI search has made link authority irrelevant.
Homepage Traffic Acts as a Trust Signal
Traffic was the second most important predictor, but only once domains reached meaningful scale.
According to Higglo’s breakdown of the dataset, websites receiving fewer than 190,000 monthly visitors showed relatively similar citation rates regardless of traffic differences.
Above that threshold, citation frequency increased substantially.
Homepage visibility was especially important. Domains with more than 7,900 monthly organic visits to their homepage were cited roughly twice as often as sites with weaker homepage performance.
The implication is difficult to ignore. AI retrieval systems appear far more likely to surface domains that already demonstrate strong visibility and trust across the wider web.
This aligns closely with what we covered earlier on why Google says AI search optimization is still SEO.
Reddit and Quora Mentions Help Smaller Brands Compete
One of the more interesting findings involved community platforms.
Domains with strong mention volumes across Reddit and Quora received up to four times more citations than domains with limited discussion activity.
For smaller brands, this may be one of the most realistic paths to AI visibility.
Building tens of thousands of referring domains is difficult. Earning mentions in active communities is considerably more achievable.
The study suggests that AI systems may treat community discussion as an entity reinforcement signal. Brands that are regularly mentioned, quoted and discussed appear more likely to be recognized as relevant sources.
The findings also support a growing theory in AI search: visibility is not just about rankings and backlinks. Brands that consistently appear in real discussions may strengthen the entity signals retrieval systems use when selecting sources.
This aligns with what we covered earlier on whether brand mentions may matter more than backlinks in AI search.
Importantly, the signal seems tied to genuine discussion rather than promotional activity. Organic mentions appear to matter far more than planted comments or artificial engagement.
Content Depth and Freshness Still Matter
Content quality signals also showed a measurable impact.
Articles exceeding 2,900 words averaged 5.1 citations, compared with 3.2 citations for articles under 800 words.
However, the study suggests that comprehensiveness matters more than word count alone.
Pages that covered multiple aspects of a topic, incorporated supporting data and provided complete answers performed better than thin long-form content.
Freshness was another notable factor.
According to Beamtrace’s review of the findings, pages updated within the previous three months averaged six citations compared to 3.6 citations for older content.
Pages using clear structure, question-based headings and concise sections also showed significantly higher citation rates.
Why This Matters for AI Search Optimization
The findings reinforce a pattern that has appeared repeatedly across recent AI search studies.
Rather than replacing SEO, AI retrieval systems appear to build on many of the same signals search engines have relied on for years.
Backlinks, authority, traffic, topical relevance and brand mentions all continue to matter.
AI search may change how users consume information, but the underlying discovery systems still appear heavily influenced by traditional web authority.
For marketers hoping GEO would replace SEO, the data points in the opposite direction.
What Did Not Work?
The study also examined LLMs.txt, a file some marketers have promoted as an AI equivalent of robots.txt.
The impact was negligible.
Removing the signal slightly improved predictive accuracy in the study’s models, suggesting it currently contributes little to citation visibility.
Review platform presence, however, did matter.
Domains appearing on platforms such as Trustpilot, G2 and Capterra averaged between 4.6 and 6.3 citations. Domains without those listings averaged just 1.8 citations.
One Limitation Worth Noting
The findings remain correlational.
Sites with 32,000 referring domains did not necessarily earn citations because of backlinks alone.
Authority, traffic, content quality and brand recognition tend to reinforce each other, making it difficult to isolate any single factor as purely causal.
Still, the scale of the dataset makes the overall direction difficult to dismiss.
For a broader context on how these signals connect, see our earlier coverage on AI citation ranking factors and what the data actually shows.
The Query Post View
The most important takeaway is not that ChatGPT cites pages ranking number one on Google.
The important takeaway is that AI search still appears deeply connected to the broader web authority ecosystem.
Backlinks, brand mentions, trusted domains and strong organic visibility continue to influence who gets seen and cited.
The AI search era may change interfaces, traffic patterns and user behavior. But based on the largest citation study published so far, it has not rewritten the fundamentals of authority.
If anything, the findings suggest that the websites winning in AI search are often the same websites that have spent years winning in traditional search.
