An increasing number of people are turning to generative AI for answers, and even Google has integrated this technology into its search engine through AI Overviews. To assist SEO professionals, and now also those involved in GEO (Generative Engine Optimization), in better understanding the operation of generative search engines and how to position themselves in their responses, the platform Rankscale.ai has analyzed the citations from various AI systems.
The report specifically analyzed nearly 8,000 citations in 57 different queries made by ChatGPT, Perplexity, Google AI Overviews, and Gemini. From this, a series of patterns have been identified, providing a more comprehensive view of the types of content most likely to be cited by each AI.
The analysis has revealed that the AI systems analyzed do not follow the same criteria when selecting which content to cite in their responses.
In the case of the GPT-4o model from ChatGPT, OpenAI’s AI prefers established, factual, and high-authority sources. The study shows that the most often cited source is Wikipedia (27% of the dataset), followed by outlets like Reuters (6%) and Financial Times (3%).
At the other end of the spectrum, user-generated content (forums and social networks), brand blogs, and product pages were hardly cited.
This data suggests that the most suitable GEO strategy for being cited by ChatGPT should focus on having a presence in reliable media, building authority, and opting for a certain level of neutrality. A good Wikipedia entry and references in leading media can make a significant difference.
For its part, Google’s Gemini 2.0 Flash model favors the combination of authoritative sources with community-generated content. Approximately 39% of its citations were from blog content and 26% from news websites.
The most cited domain was YouTube (~3%). Mentions of Wikipedia were also found, but they were more scarce. Community content represented approximately 2%.
To enhance your presence in Gemini’s results, focus on enhancing the depth of your content and achieving broad web coverage. Authority, targeting quality content, and having influence on YouTube will also be relevant factors.
Approximately 38% of its citations come from editorial blogs, while about 23% correspond to news and 7% to product blogs. Domains of specialized review websites like NerdWallet, Consumer Reports, or Investopedia play a significant role.
Although user-generated content is occasionally mentioned, Perplexity usually avoids low-quality sources. To ensure this AI considers and cites you, it is necessary to create specialized and useful content, base it on verified data, and demonstrate expertise. Another good strategy is to encourage conversation in forums.
The AI-generated views from Google show a greater diversity in their sources. The content with the highest presence includes blog-style articles (approximately 46%) and news (~20%). Community content, product blogs, and social media also play a relevant role.
While Reddit was the most cited site, YouTube and Quora also had a strong presence, with Wikipedia yielding more modest results.
To support GEO aimed at appearing among the results of AI Overview, it will be important to focus on deep and multifaceted content on your site. Detailed guides, comparisons, and lists are highly valued. It is also advisable to be present in active communities like Reddit or Quora, but also in media outlets and question-and-answer websites.
There are also differences based on the type of search.
For B2B queries, more professional sources stand out: company blogs (~17%), directories like Clutch.co, analyst reports (Gartner, Statista), and niche publications (TechTarget, Gartner). LinkedIn also holds weight.
Publishing detailed content on your own blog, creating product comparisons and enriched reports will be key to positioning yourself among the results of generative search engines. Additionally, it will be important to build your presence in specialized media and directories.
The most cited content comes from popular media (Forbes, Business Insider), product reviews, Wikipedia, YouTube, technology websites (PCMag, CNET) and communities like TripAdvisor or Quora. Brand websites or blogs are rarely directly cited.
An appropriate strategy would be to focus on obtaining mentions on review sites, technology/lifestyle blogs, community forums, and Wikipedia.
In this case, generative search engines cite objective content such as studies, government reports, specialized press, and even industry-relevant awards. The focus is more neutral and factual. News sources and blogs represented nearly 70% of the citations, while user opinions and direct provider promotions had little weight.
It is recommended to create objective content based on data and ensure mentions in market reports, sector rankings, and reputable media.
The analysis demonstrates that a solid organic presence remains the foundation. AI systems primarily rely on the best-positioned pages in Google to configure their responses.
Therefore, if your brand is capable of ranking well and meeting the necessary SEO criteria (quality content, site authority, EEAT standards, etc.), it will be closer to being cited by generative search engines.
However, it is not all about occupying the first position, as you can also be cited even if you are not at the top, as long as your content has quality and authority. The key is to generate real value from multiple sources, not just from your own blog.
Another key piece of data extracted from the analysis was which brands are mentioned most frequently by the AIs. On average, ChatGPT and AI Overviews mention the fewest brands, with an average of 3-4 brands per response, and the ones referenced are usually leading brands like Netflix, Expedia, or Salesforce.
Meanwhile, Perplexity cites up to 13, citing leading brands, but also giving space to niche brands. As for Gemini, it falls in an intermediate range, citing an average of about 8 brands, including both leading and secondary brands.
Generally, the most cited brands were established leaders in their market like Salesforce for CRM, Netflix for streaming, Apple and Samsung for smartphones, or Nike for sportswear.
Photo: GPT-4o
Your email address will not be published. Required fields are marked *
Δ