A new large-scale analysis of how ChatGPT sources information for its answers has uncovered a critical link between traditional search engine optimization and success in the emerging world of AI conversational tools. The study, conducted by AirOps, examined over 15,000 initial prompts and the subsequent web pages processed by the AI, revealing that securing the number one position on a Google search results page multiplies the probability of being cited in a ChatGPT response by 3.5 times.
The Hidden Selection Process Behind AI Citations
While ChatGPT has become a daily tool for millions, the mechanisms guiding its response generation remain largely opaque to users and publishers alike. The AirOps study sheds light on a fundamental aspect: ChatGPT does not cite every piece of content it consults. In fact, the AI performs a rigorous internal filtering process. Out of a total of 548,534 web pages that ChatGPT read or processed during the study, only approximately 82,000—roughly 15%—were ultimately presented as citations to the user. This means a staggering 85% of the sources consulted by the AI never appear as a visible reference.
Citation Rates Vary by Query Intent
The research also highlighted significant differences in citation behavior based on the type of user prompt. Queries related to product discovery and tutorials enjoyed the highest citation rates, at 18.3% and 16.9% respectively. Other categories, such as validation and comparison searches, resulted in fewer direct citations, suggesting the AI’s sourcing strategy adapts to the intent behind the question.
Key Factors That Influence ChatGPT’s Citation Choices
Delving deeper, the analysis pinpointed two primary factors that significantly increase a webpage’s chances of transitioning from a consulted source to a cited one: semantic coherence and search engine ranking.
The Power of Title-Query Alignment
A clear, direct relationship between the page’s title and the user’s query emerged as a powerful predictor. Pages where the title overlapped with the search query by 50% or more were cited more than twice as often as those with lower coherence. Furthermore, readability played a crucial role. Pages with higher Flesch Reading Ease scores—indicating simpler sentence structures and shorter paragraphs—were consistently favored, aligning with findings that characterize ChatGPT as an “impatient reader” more likely to extract information from the opening sections of content.
SEO Success Directly Translates to AI Visibility
The most striking finding, however, concerns the direct correlation between Google rankings and ChatGPT citations. The data shows that pages occupying the coveted first position in Google’s Search Engine Results Pages (SERPs) are cited 3.5 times more frequently than pages that do not rank within the top 20 results. Specifically, 43.2% of all pages holding the number one spot in Google were cited by ChatGPT. The study’s authors note, “When comparing cited pages with the top 20 Google results for the original and derived queries, we found that 55.8% of all cited pages ranked in the top 20 for at least one query.” This establishes that strong organic search performance not only increases the likelihood of entering ChatGPT’s retrieval set but also dramatically boosts the odds of being selected for the final citation.
Domain Authority Plays a Surprising, Nuanced Role
Contrary to what one might expect, the study revealed that sheer domain authority (DA) is not the ultimate gatekeeper for ChatGPT citations. A significant 74% of all citations went to websites with a domain authority score below 80. The majority of citations were concentrated on sites in the mid-range of authority, with a DA between 20 and 80, accounting for 63.6% of all citations. Intriguingly, the group of domains with a low DA (20-40) actually had slightly more representation than the highest-authority sites (DA 80-100), at 26.0% versus 25.4%.
The “Fan-Out” Phenomenon and Niche Expertise
This counterintuitive result is explained by what the researchers term the “fan-out” phenomenon. When processing a complex query, ChatGPT often decomposes it into several more specific, niche internal searches. In these narrow, intent-driven explorations, websites with medium authority but highly specialized, expert content can outperform larger, generalist domains with higher overall authority. The opportunity lies not in outranking publishing giants everywhere, but in competing effectively within the specific retrieval environments ChatGPT uses to build its answers.
Strategic Implications for Content Creators
The authors conclude that brands should focus on generating sufficient topical authority, precise query alignment, and comprehensive coverage to appear in these retrieval environments. The clearest opportunities often reside in long-tail topics, where creators can develop and consistently update content that addresses the precise questions and subtopics ChatGPT explores when crafting a response. This approach aligns with previous research recommending the use of direct-question H2 and H3 headers and structured data markup, as pages employing such technical mark-up are 1.7 times more likely to be cited.
As AI tools like ChatGPT increasingly become primary gateways to information, understanding their sourcing algorithms is paramount for digital visibility. This research confirms that foundational SEO principles—clarity, relevance, and user-centric content structure—remain vital. However, it also unveils a new competitive landscape where niche expertise and strategic alignment with AI retrieval patterns can allow websites of all sizes to earn valuable citations and the qualified traffic that accompanies them.
