Only 12% of the sources mentioned by AI make it into Google’s top 10

In 2025, the convergence between artificial intelligence and SEO poses a major challenge for web players. While natural SEO has long dictated content visibility, the way generative AI selects its sources could well disrupt this established order. A recent study highlights that, on average, only 12% of links cited by assistants like ChatGPT or Perplexity appear in Google’s top 10 for a given query. This discrepancy can be explained by the divergent criteria that govern these two worlds: while Google often prioritizes popularity and authority, AI focuses more on regional or contextual synthesis. The direct consequence? The visibility of content no longer depends solely on its position on a search engine, but also on its ability to feed these AI systems, which construct a quasi-autonomous response, without necessarily referencing the first pages. To understand the issues, we must therefore analyze how these sources are selected, what this implies for SEO, and how to better control its presence in this new situation. Understanding the Logic of Source Selection in AI in 2025

AI assistants don’t just index the web like Google did in its early days. Their operation relies on sophisticated algorithms that merge a multitude of query variants to provide a synthetic response, often without explicit reference. Source selection is based on several criteria: contextual relevance, perceived reliability, and the ability to cover all the issues related to the query. But these criteria differ greatly from those of traditional search engines.

To illustrate this, let’s take the example of a query about the latest geopolitical crisis or a pandemic. Depending on the type of AI, the preferred source could be a report from a specialized agency or an analysis from a poorly referenced media outlet, rather than a Google News article or a publication from Le Monde. One of the major challenges lies in the fact that these sources, often less visible in traditional SERPs, sometimes contain very high-quality information but remain underutilized by the majority of traditional SEO strategies.

Algorithms are now moving towards a logic of “expanded contextualization.” This means that instead of answering the exact query, the AI will generate a response by combining different variations of the same theme. This, in practice, favors deep sources, which are less visible but highly relevant in certain specific cases. The dominant sources in the AI sphere then become those that have been able to adapt to this multiple logic, often by using sophisticated technical language or offering highly specialized content. List of key criteria for AI source selection

🔍 Contextual relevance: AI prioritizes sources that provide precise insights tailored to the specific query.

🔗 Reliability and authority: The credibility of sources, such as Wikipedia or institutional websites, often influences the selection.

• 💡 Diversity of viewpoints: To avoid bias, AI blends different perspectives from various sources.
• 🌐 Accessibility: Ease of access to information quickly, particularly via APIs or databases, also plays a role.
• 📈 Impact of recent or viral content: Sometimes, breaking news or virality can change the outcome of source selection.
• This process partly explains the weak correlation between sources cited by AI and the results at the top of traditional SERPs. The problem isn’t so much the quality of the sources, but the fact that their use isn’t based on an initial ranking logic, but on adaptation to the conversational context.
• Impacts of this gap on traditional web SEO

The gap between AI sources and Google’s top 10 is disrupting SEO strategy. Indeed, the majority of content ranked in the top 10 will not necessarily be picked up or cited by these new intelligences. This calls into question the ability of companies to position themselves simply by aiming for the top spots in Google in the hope of appearing in AI responses.

→ À lire aussi Discover how to optimize your visibility with AI in the era of generative engines like ChatGPT and Perplexity Local SEO · 03 Aug 2025

According to a study conducted by

Ecommerce Nation

, 76% of the sources used in AI Overviews come from the Google top 10, but the majority of content cited by ChatGPT or Gemini does not come from this selection. This shows that it is becoming crucial to adopt an approach focused on depth and relevance, rather than visibility alone, in traditional SEO. New strategies to adopt 🛠️ Produce rich, structured content that’s essential for in-depth research.🔎 Build site authority by multiplying credible references and diversifying sources (Wikipedia, Reddit, YouTube).

🎯 Target long-tail queries and variants to cover a wider range of intent.

• 📊 Maintain constant monitoring to identify sources favored by AI using tools like Ahrefs or SEMrush.
• 📝 Create engaging content with fresh perspectives to attract the attention of these summarization systems.
• Discover various reliable and relevant sources of information to deepen your knowledge on the subject. Find useful references, articles, documents, and resources here to learn more.
• What the weak link between sources and the top Google rankings means for the future of SEO
• SEO professionals must rethink their approach. The race for the top position in Google no longer guarantees ultimate visibility in the new AI landscape. The priority must now be the production of high-quality content, geared towards long-tail queries, and in formats likely to be exploited by these intelligent assistants.

→ À lire aussi ideal number of words for optimizing an ecommerce category page Local SEO · 18 Aug 2025

A detailed analysis of these new trends shows that focusing solely on position in Google is a risky strategy in the years to come. Frequently Asked Questions About the Influence of AI Sources and SEO

Why don’t sources cited by AI always appear in Google’s top 10?

AI algorithms play a different role than Google. They prioritize contextual relevance and the diversity of sources scattered across often unknown databases, which explains why their citations don’t always rank high in SERPs. How can we improve visibility in generative AI systems? It’s necessary to produce in-depth and diverse content, integrating credible sources and targeting specific, long-form queries. Creating rich content helps attract the attention of AI algorithms for better integration into their responses.

Does traditional SEO still have a future?

Yes, but it must evolve to align with these new logics. SEO is no longer limited to simple ranking on Google; it must also focus on semantics, the depth of content, and its ability to be cited by AI in a specific context.

What are the main risks for SEO in the face of this shift?

Limiting oneself to ranking in Google could become a strategic error. The difficulty lies in knowing the sources favored by AI, which are not always the most visible in traditional results.

Écrit par

Kevin Grillot

Consultant Webmarketing & Expert SEO.

Voir tous les articles →