The specialist media Search Engine Land has taken a closer look at the content cited by generative AI. Discover the findings of its study.
Generative AI web search, which has become more widespread since the launch of ChatGPT Search, is challenging certainties in the world of SEO. Little by little, Google Search seems to no longer be the alpha and omega of SEO. The ability to appear in the responses of conversational agents is therefore becoming a new skill to develop, but this discipline remains unclear: few indicators today allow us to clearly understand what types of content appeal to these artificial intelligences.
In a study published this Monday, May 12, 2025, the specialist media Search Engine Land attempted to unravel this mystery by analyzing 8,000 AI citations obtained using data from the Rankscale.ai tool. Here’s an overview.
The main sources of ChatGPT
Search Engine Land reports that ChatGPT “strongly favors established, authoritative, and factual sources,” particularly encyclopedias and major news outlets. Wikipedia dominates the results: the collaborative encyclopedia, already a strong performer on Google Search, accounts for 27% of the citations observed. In the United States, the Reuters news agency and the Financial Times also appear to be in the OpenAI chatbot’s good books. This comes as no surprise: these two news outlets are among ChatGPT Search’s partners.
Conversely, Search Engine Land notes that “user-generated content (UGC), such as forums and social media, is virtually absent,” as are blogs or vendor product pages. The American media outlet offers some recommendations to improve your visibility on ChatGPT:
- Build Authority:Â Establishing strong credibility is essential to being cited by ChatGPT.
- Visibility in neutral sources:Â Having a presence on Wikipedia and being mentioned in reliable sources such as specialist blogs and news media will be beneficial.
- Preference for non-commercial sources:Â Language models like ChatGPT favor neutral sources of information over purely commercial sites.
Quotes from Gemini, Perplexity, AI Overviews
What about ChatGPT’s competitors? The study makes some observations regarding Gemini, Perplexity, and AI Overviews.
Gemini: more diverse quotes than ChatGPT
The Google-developed chatbot also cites Wikipedia, but “much less prominently .” Indeed, Gemini tends to diversify its sources: 39% of citations come from blogs, 26% from news sites, and 2% from community content. Search Engine Land considers that Gemini “seems adept at blending professional opinions with peer commentary, particularly for consumer questions. ”
Perplexity: Reliable Sources and Expert Opinions
For its part, Perplexity seems to emphasize expertise and freshness of information. 38% of citations come from blogs, 23% from news sites and 9% come from “domains known for their expert opinions” . UGC, while not representing the majority of citations, is still more used than on ChatGPT. Thus, to ensure a presence on Perplexity, Search Engine Land advises “cultivating your presence on high-authority niche sites and on respected review platforms relevant to your sector”.
AI Overviews: More Community and Branded Content
AI Overviews, which is still inaccessible in France, is a “broad aggregator .” The overviews cite blogs (46%) and mainstream news sites (20%), but also leave room for community content—sites like Reddit and Quora represent 4% of citations—and even product blogs written by vendors (7%). Therefore, for Search Engine Land, “a multifaceted web presence” will be a key element in ensuring citations in AI Overviews.
B2B vs. B2C: Differences in Citations
Overall, Search Engine Land highlights a significant difference between citations that respond to B2B queries and those that address B2C queries. B2C queries, such as “best smartphone brands,” result in citations from popular media, review sites, forums, or Wikipedia. They combine expertise with public opinion, reflecting consumer decision-making. Official websites have a very low presence.
For B2B queries, such as “best CRM software,” search engines primarily cite company blogs, directories, trade publications, or analyst reports. The emphasis is on expertise, accurate data, and professional credibility.
Mixed-interest queries, such as “renewable energy companies,” rely primarily on neutral sources, such as public reports, official rankings, or academic publications. They aim to inform a broad audience with reliable and objective data, while limiting the amount of reviews or promotional content.
AI and SEO: Some general recommendations
In conclusion, the study presents some good practices to adopt to maximize your chances of being cited by AI engines:
- Be present on authoritative third-party sites:Â Target recognized blogs (including niche ones), quality media, and influential review platforms in your industry.
- Strengthen your base authority:Â Take care of your Wikipedia page and your Google Knowledge Panel, two key signals for AIs like ChatGPT.
- Create expert and structured content:Â produce comparative guides or detailed analyses on your site.
- Participate in online communities:Â Reddit, Quora, and niche forums are helpful, especially for B2C queries.
- Strengthen EEAT signals:Â Just like on Google Search, it’s important to update your content, cite your sources, showcase your expertise, and collect authentic reviews.
