SE Rating analyzed 129,000 distinctive domains throughout 216,524 pages in 20 niches to determine which components correlate with ChatGPT citations.
The variety of referring domains ranked as the one strongest predictor of quotation chance.
What The Knowledge Says
Backlinks And Belief Indicators
Hyperlink range confirmed the clearest correlation with citations. Websites with as much as 2,500 referring domains averaged 1.6 to 1.8 citations. These with over 350,000 referring domains averaged 8.4 citations.
The researchers recognized a threshold impact at 32,000 referring domains. At that time, citations almost doubled from 2.9 to five.6.
Area Belief scores adopted an identical sample. Websites with Area Belief under 43 averaged 1.6 citations. The advantages accelerated considerably on the prime finish: websites scoring 91–96 averaged 6 citations, whereas these scoring 97–100 averaged 8.4.
Web page Belief mattered lower than domain-level alerts. Any web page with a Web page Belief rating of 28 or above obtained roughly the identical quotation fee (8.3 common), suggesting ChatGPT weighs total area authority extra closely than particular person web page metrics .
One notable discovering: .gov and .edu domains didn’t mechanically outperform industrial websites. Authorities and academic domains averaged 3.2 citations, in comparison with 4.0 for websites with out trusted zone designations.
The authors wrote:
“What in the end issues just isn’t the area title itself, however the high quality of the content material and the worth it offers.”
Visitors & Google Rankings
Area site visitors ranked because the second most vital issue, although the correlation solely appeared at excessive site visitors ranges.
Websites underneath 190,000 month-to-month guests averaged 2 to 2.9 citations no matter precise site visitors quantity. A website receiving 20 natural guests carried out equally to 1 receiving 20,000.
Solely after crossing 190,000 month-to-month guests did site visitors correlate with elevated citations. Domains with over 10 million guests averaged 8.5 citations.
Homepage site visitors particularly mattered. Websites with a minimum of 7,900 natural guests to their foremost web page confirmed the best quotation charges.
Common Google rating place additionally tracked with ChatGPT citations. Pages rating between positions 1 and 45 averaged 5 citations. These rating 64 to 75 averaged 3.1.
The authors famous:
“Whereas this doesn’t show that ChatGPT depends on Google’s index, it suggests each techniques consider authority and content material high quality equally.”
Content material Depth & Construction
Content material size confirmed constant correlation. Articles underneath 800 phrases averaged 3.2 citations. These over 2,900 phrases averaged 5.1.
Construction mattered past uncooked phrase depend. Pages with part lengths of 120 to 180 phrases between headings carried out finest, averaging 4.6 citations. Extraordinarily brief sections underneath 50 phrases averaged 2.7 citations.
Pages with knowledgeable quotes averaged 4.1 citations versus 2.4 for these with out. Content material with 19 or extra statistical information factors averaged 5.4 citations, in comparison with 2.8 for pages with minimal information.
Content material freshness produced one of many clearer findings. Pages up to date inside three months averaged 6 citations. Outdated content material averaged 3.6.
Surprisingly, the uncooked information confirmed that pages with FAQ sections really obtained fewer citations (3.8) than these with out (4.1). Nonetheless, the researchers famous that their predictive mannequin considered the absence of an FAQ part as a destructive sign. They recommend this discrepancy exists as a result of FAQs usually seem on less complicated assist pages that naturally earn fewer citations.
The report additionally discovered that utilizing question-style headings (e.g., as H1s or H2s) underperformed easy headings, incomes 3.4 citations versus 4.3. This contradicts commonplace voice search optimization recommendation, suggesting AI fashions might favor direct topical labeling over query codecs.
Social Indicators & Overview Platforms
Model mentions on dialogue platforms confirmed sturdy correlation with citations.
Domains with minimal Quora presence (as much as 33 mentions) averaged 1.7 citations. Heavy Quora presence (6.6 million mentions) corresponded to 7.0 citations.
Reddit confirmed related patterns. Domains with over 10 million mentions averaged 7 citations, in comparison with 1.8 for these with minimal exercise.
The authors positioned this as significantly related for smaller websites:
“For smaller, less-established web sites, participating on Quora and Reddit presents a approach to construct authority and earn belief from ChatGPT, much like what bigger domains obtain by backlinks and excessive site visitors.”
Presence on evaluation platforms like Trustpilot, G2, Capterra, Sitejabber, and Yelp additionally correlated with elevated citations. Domains listed on a number of evaluation platforms earned 4.6 to six.3 citations on common. These absent from such platforms averaged 1.8.
Technical Efficiency
Web page pace metrics correlated with quotation chance.
Pages with First Contentful Paint underneath 0.4 seconds averaged 6.7 citations. Slower pages (over 1.13 seconds) averaged 2.1.
Velocity Index confirmed related patterns. Websites with indices under 1.14 seconds carried out reliably properly. These above 2.2 seconds skilled steep decline.
One counterintuitive discovering: pages with the quickest Interplay to Subsequent Paint scores (underneath 0.4 seconds) really obtained fewer citations (1.6 common) than these with reasonable INP scores (0.8 to 1.0 seconds, averaging 4.5 citations). The researchers advised very simple or static pages might not sign the depth ChatGPT seems for in authoritative sources.
URL & Title Optimization
The report discovered that broad, topic-describing URLs outperformed keyword-optimized ones.
Pages with low semantic relevance between URL and goal key phrase (0.00 to 0.57 vary) averaged 6.4 citations. These with highest semantic relevance (0.84 to 1.00) averaged solely 2.7 citations.
Titles adopted the identical sample. Titles with low key phrase matching averaged 5.9 citations. Extremely keyword-optimized titles averaged 2.8.
The researchers concluded: “ChatGPT prefers URLs that clearly describe the general matter fairly than these strictly optimized for a single key phrase.”
Components That Underperformed
A number of generally beneficial AI optimization techniques confirmed minimal or destructive correlation with citations.
FAQ schema markup underperformed. Pages with FAQ schema averaged 3.6 citations. Pages with out averaged 4.2.
LLMs.txt recordsdata confirmed negligible impression. Outbound hyperlinks to high-authority websites additionally confirmed minimal impact on quotation chance.
Why This Issues
The findings recommend your present search engine marketing technique might already serve AI visibility targets. When you’re constructing referring domains, incomes site visitors, sustaining quick pages, and maintaining content material up to date, you’re addressing the components this report recognized as most predictive.
For smaller websites with out in depth backlink profiles, the analysis factors to group engagement on Reddit and Quora as a viable path to constructing authority alerts The information additionally suggests specializing in content material depth over key phrase density.
The researchers notice that components are interdependent. Optimizing one sign whereas ignoring others reduces total effectiveness.
Trying Forward
SE Rating analyzed ChatGPT particularly. Different AI techniques might weight components in a different way.
SE Rating doesn’t specify which ChatGPT model or timeframe the info represents, so these patterns needs to be handled as directional correlations fairly than proof of how ChatGPT’s rating algorithm works.
Featured Picture: BongkarnGraphic/Shuttersrtock
