LLM and SEO: AI Redefines the Rules of Search Engine Optimization

AI Tools: A New Form of Search Engine

Jessica Michenaud, SEO project manager, asserts that large language models (LLMs) are emerging as a new form of search engine. She emphasizes the importance of these tools for brand visibility. According to a study conducted by Ipsos, about 39% of French people use generative AI tools, and among them, 48% use them for conducting searches. These figures highlight the growing significance of these technologies in users' daily lives.

For Michenaud, it is crucial to focus on LLM SEO. She explains that for a brand to be visible in the responses provided by these tools, it must be well-optimized. "Being well-optimized by these tools means ensuring that your brand appears in the answers users receive," she specifies. Thus, LLMs are becoming an essential visibility channel for brands, complementing traditional SEO strategies.

How to Write for LLMs

The good news, according to Jessica Michenaud, is that writing for LLMs does not require reinventing one's writing style. She stresses the importance of structuring content rather than changing style. "It's not about changing your writing style, but about better structuring your content," she explains. This distinction is crucial, as the topic can generate concerns within the industry.

Michenaud identifies several levers to optimize content for LLMs: directly answering the question at the beginning of the content, varying response formats (text, lists, tables) to make information more easily extractable, and ensuring clarity and overall readability of the content. Well-structured content is more easily understood by AIs, which also enhances the user experience.

However, she reminds us of an essential prerequisite: without basic SEO, there is no optimization in LLMs. "LLMs rely on what is already indexed and visible on the web. SEO remains the foundation," she insists.

"Humans Must Remain at the Heart of Writing"

Although structuring content for LLMs is important, Michenaud emphasizes that this should not come at the expense of the primary intention of the content, which is to be useful for humans. She asserts that optimization should never take precedence over intent. Content that does not answer a genuine human question has no reason to exist, whether it is readable by AIs or not.

She adds that LLMs have been trained on ordinary content produced by humans, and there is no comprehension issue for AIs when faced with well-written and natural content. "There is no comprehension issue for AIs when faced with well-written and natural content," she reminds us. Thus, there is no need to make writing contortions to be readable by the models.

Quality will always take precedence over technical optimizations. Content that truly answers the user's question, with real added value, will be appreciated by both Google and LLMs. This is not a revolution: it is the basic editorial requirement. Clear, structured, and quality content satisfies both Google and LLMs. No parallel strategy, no double writing. "Yes, LLMs matter, as users are turning to them more and more. But no, traditional SEO is not dead. It remains the foundation without which no visibility is possible, neither on Google nor in AI responses," she concludes.