GEO: How AI is Redefining Online Visibility

The Evolution of SEO in the Age of AI

Traditional SEO, firmly rooted in the digital landscape, is seeing its dynamics shift with the emergence of summaries produced by artificial intelligences. Generative Engine Optimization (GEO), a new approach, aims to ensure visibility in the responses generated by these AI models. As conversational interfaces tend to replace classic search engines, the battle for visibility is moving to these new platforms.

Historically, SEO has been the compass for content creators, dictating the rules for appearing at the top of search results. However, with the rise of GEO, this strategy must adapt to include AI responses, which are becoming a new playing field for brands and content creators.

Citability: A Determining Criterion

Unlike traditional search engines, generative AIs do not always explicitly select sources. They do not browse the web to find specific pages but generate responses based on the patterns they have learned. In this context, the notion of source becomes blurred.

AI responses rely on hybrid criteria such as semantic relevance, content freshness, perceived credibility, and coherence among multiple sources to avoid hallucinations. Traditional SEO signals, like domain authority or backlinks, continue to influence, but their impact is diminished.

To maximize the chances of being referenced by an AI, content must be structured in a way that allows for easy extraction and rephrasing. Elements such as clear tables, precise FAQs, sourced definitions, and numbered lists are particularly effective. Thus, a modest page in terms of SEO can outperform a more authoritative competitor if it is better suited for automatic synthesis.

Most effective tools now integrate a retrieval module, or a browser version like ChatGPT Atlas. The AI queries an updated index, retrieves relevant documents, assesses their coherence, and reformulates a summary. This logic resembles that of a document retrieval system: relevance, credibility, freshness, and coherence among sources become essential criteria.

Alain Guisado from Linksgarden shared an internal experience where they managed to manipulate ChatGPT into claiming that the favorite pizza of the French in 2025 would be Hawaiian pizza. This anecdote illustrates how information can be influenced by well-thought-out strategies.

The Levers of GEO and Their Limits

Experts identify three main levers to influence GEO and maximize the likelihood of appearing in AI responses. First, it is crucial to be present in the corpora consulted by the models, which implies a presence in recognized media, specialized sites, or structured databases.

Second, it is necessary to increase credibility signals, ensuring that the domain is recognized and that thematic expertise is clear. Finally, maximizing the probability of being cited requires producing highly structured content, such as definitions, sourced data, or comparisons.

However, some more aggressive strategies may be attempted, such as artificially multiplying evidence of the existence of information or producing extremely "summarizable" content. These approaches may be effective in the short term, but they carry risks, especially if they become too visible and counterproductive.

Biases and Risks to Reliability

The possibility of influencing AI citations raises questions about the overall reliability of information. This phenomenon recalls the excesses of black hat SEO, but in the context of AI. Generated responses can exhibit several types of biases, such as availability bias, where the most prevalent content dominates the responses, even if they are not the most reliable.

Other biases concern the sources consulted, particularly if the research relies on dominant ecosystems, such as English-language media. Marketing bias can also favor content optimized for synthesis at the expense of deeper but less structured work.

These biases can amplify mediocre content, misinform on sensitive topics, and create artificial information bubbles. In the long run, this could lead to a widespread loss of trust if responses appear too manipulated.

The Economic Interest of GEO

The rise of GEO is largely explained by its economic interest. AI-generated responses become a new distribution surface for brands, allowing them to capture leads, enhance their visibility, or influence certain decisions. In sectors like B2B, finance, health, or education, being cited by an AI can steer significant choices.

However, these practices carry risks. If the best-optimized content dominates, it can amplify weak or inaccurate information. Artificial information bubbles may emerge, where a small group of optimized sites loops in AI citations.

Towards Safeguards for GEO

To mitigate these risks, several protective mechanisms are developing. AI systems are gradually reinforcing the weighting of reliability, relying on signals close to the E-E-A-T (Experience, Expertise, Authority, Trustworthiness) concept to evaluate sources. Architectures are also seeking to diversify sources to avoid monoculture effects.

Methods for detecting manipulation are emerging, allowing models to identify networks of suspicious sites or nearly identical content. Some platforms are also developing automated fact-checking systems to compare responses across multiple sources and limit errors.

The question of traceability is becoming central, with projects aimed at explaining why a source is cited and allowing for contestation. At the regulatory level, the European AI Act could play a role in strengthening transparency obligations and risk management.

While these regulations should not eliminate GEO, they could encourage more qualitative and transparent practices. Alain Guisado reminds us that AI should be seen as a synthesis tool and a research assistant, provided that access to sources, the ability to verify, and solid safeguards are maintained.