GEO: how to optimise your content for generative AI

With the rise of generative AI, the rules of online visibility are changing rapidly. The GEO (Generative Engine Optimization) is emerging as a new marketing lever, but one that still has a long way to go. Is it time to review your content strategy? And how can you make yourself more visible in AI-generated responses than in a simple list of links? Here's how.

If you work in digital marketing, you've probably heard of GEO, for Generative Engine Optimization. Behind this acronym lies an evolution in usage. Some Internet users no longer click first on a list of links, but consult a summary response produced by an AI (ChatGPT, Claude, Gemini, etc.). This development is leading to a major developments in search engine optimisation and SEO Content optimisation: optimising content not just to appear in the search results (SERPs) of Google and other search engines, but also to increase the chances of content being picked up, quoted or used in responses generated by AIs.

The term has become popular since 2023, driven by the democratisation of AI chatbots. Today, the overwhelming majority of SEO agencies and consultants offer this service to their clients, presenting it as an essential aspect of any 360° web visibility strategy.

Why is GEO becoming a strategic marketing issue?

AI chatbots have reshaped the rules of the web by transforming the way users search for information. More and more Internet users are using them as search engines in their own right to obtain information, compare solutions and even guide their purchasing decisions. According to a recent study, 62 % of users consult AI before making a purchase for information, comparisons and recommendations.

Web traffic from AI engines is still lower than Google's: 0.25 % of total traffic, according to a study by Ahrefs. But Ahrefs also points out that AI has become their best performing channel in terms of conversion rate (10 %). Indeed, users who have carried out targeted searches show a higher level of engagement. What's more, the share of web traffic coming from AI is constantly climbing. This is driving more and more organisations to integrate the major language models (LLM) to their visibility strategy. In fact, almost half of Internet users say that lists of sites or products generated by AI influence their purchasing decisions. This represents a significant market share. OpenAI (ChatGPT), for example, reports 900 million weekly visitors on its platform.

Another reason behind the GEO boom is the launch in 2024 of’AI Overviews, a Google feature that summarises SERP results to answer web users' questions. Due to a more restrictive regulatory framework, this functionality is not yet available in France. If it were to be deployed, it would only accentuate the need to optimise content. AI Overviews always displays a list of sources next to its summaries. Brands that want to remain visible in the age of AI would do well to be on this list.

As a new discipline, GEO is constantly evolving. While marketing experts have established a number of basic rules, these are still subject to development by LLMs and need to be validated by regular research and testing. Keep an open mind, experiment and let your own results guide your approach.

Towards a structural fall in the click-through rate

One of the most structuring effects of the rise of AI engines is often underestimated: lower click-through rates. By offering complete and synthetic answers, AIs reduce the need to consult sources. The result: the more users go through response engines, the less they click. This phenomenon, which has already begun with featured snippets by Google, is growing with generative AI. Should we see this as a threat? Not necessarily, because the logic is changing. We're losing click volume, but we're gaining in quality of commitment.

Content cited by an AI often comes at the decision-making stage, with a more mature intention and in a context of increased trust. For marketers, the KPI is changing: less traffic, but more impact.

GEO vs SEO: a change in logic, not a break with the past

Here is an overview of the main differences between these two practices:

	SEO	GEO
Target	Traditional search engines (Google, Bing)	LLMs such as ChatGPT, Claude, Gemini and Perplexity
Objective	Appear in the top SERP results	Being quoted in the responses of AI engines
Content	Integration of keywords, particularly long-tail keywords	Integration of conversational language reproducing the prompts entered in chatbots
Success indicators	Click-through rate Keyword ranking Time spent on page	Frequency of citations in LLMs Share of Voice*

*Share of Voice A brand's share of visibility compared to its competitors on a given channel. It reflects its presence on the web (conversations, media) and indicates its ability to capture the market's attention.

Google and LLMs do not search for information in the same way.

• Google indexes the web and returns links classified by relevance.
• The aim of LLMs is to generate an answer to a given question (a prompt) by cross-referencing several sources. At the same time, they will often include external links and/or brand names so that users can trace the source. This is where GEO comes in.

Visibility no longer depends solely on ranking, but on the ability to be selected as a reliable source.

According to a study by the universities of Guangzhou and Rutgers Comparing Google results with those of LLMs, while chatbots generally cite the top SERP results, they can also mention smaller sites if they are relevant. By offering high-quality content, you have the opportunity to stand out from the big brands.

However, the SEO and GEO are not in opposition because the fundamentals remain close. Both aim to satisfy the user's search intent and follow the E-E-A-T principle (Experience, Expertise, Authority, Reliability).

As with SEO, the more reliable and rigorous your content, the better your chances of appearing in the results.

Some specialists see GEO not as a discipline in its own right, but as a natural extension of SEO.

How to optimise your content for AI

1. Talk like your users

Queries to AI interfaces are often formulated in natural language. Work on your content based on the real questions asked by your customers, prospects and readers. To guide your content effectively, find out how your users formulate their queries:

• What questions would they ask themselves before coming to your product or service?
• What would their prompts look like?

Use tools such as Google Search Console to identify web users' queries. Examine your long tail keywords and rephrase them in the form of questions.

Conversations on LinkedIn can also provide valuable information about how your target audience expresses itself. And don't overlook internal data (FAQs, forums, customer tickets) to understand the behaviour of your own customers.

2. Treat a subject in depth, not on the surface

Google documents the use of a «query fan-out» technique in AI Overviews and AI Mode. The AI divides the question into sub-topics and performs simultaneous searches on each of them across several data sources. It then compiles these results to provide an easy-to-understand answer. Editorial consequence: an isolated, thin page is less likely to be useful than a coherent set of content covering the subject, its variants, comparisons, use cases and limitations.

For example, the prompt «What is a hybrid mattress?» could generate the following subqueries:

• What are the best brands of hybrid mattresses?
• What is the difference between a hybrid mattress and a foam mattress?
• Which hybrid mattress should you choose depending on your sleeping position?

By processing the original prompt, LLMs will create a complete summary that answers all these questions. As sources, they will therefore favour sites that answer as many sub-queries as possible.

What does this mean in practical terms?

To be visible, cover a subject exhaustively. If you sell hybrid mattresses, make sure that the angles above are all covered by articles linked together via internal linking. This shows LLMs that you understand the context. You can also include these questions in FAQs.

3. Strengthening credibility through evidence

LLMs prefer sourced content attributed to specialists or authoritative sites in their field.

Here's an example of an online shop offering guided meditations. The shop also publishes blog articles on the subject. It's in the writers' interests to include the following content in their articles:

• customer testimonials
• interviews
• quotes from experts, studies or serious sources
• links to recognised medical publications
• explicit methodologies
• identified perpetrators
• real, quantified or specific examples (company name, date, etc.), etc.

Original content is particularly valued They are the ones most likely to be taken over by AIs.

• in-house surveys
• studies conducted with a partner institute
• creative data visuals or genuinely informative tables
• insights from your own experience in the field

Unprecedented figures and unique perspectives are a magnet for great language models.

4. Keeping your content up to date

The more recent your content is and the more up-to-date it reflects the latest developments in the sector, the more likely it is to be cited. Don't forget to update them by making the necessary changes and/or updating the date in the title. A word of caution: if you change the date without enriching the content sufficiently, it may be penalised by Google and the AIs.

5. Monitor your visibility in AI

GEO introduces new indicators. Measure your progress by according to the volume of traffic generated by AI engines, as well as frequency with which your brand is mentioned. SEO tool Semrush has launched an AI Visibility Toolkit, and a number of specialist GEO tools have emerged in recent years, including OtterlyAI and Conductor.

What if GEO became vocal?

GEO is still largely associated with text-based interfaces. And yet, uses are rapidly evolving towards voice.

Optimising for the ear, not just the eye

Traditional SEO is designed for quick reading (titles, lists, keywords). Vocal GEO imposes a different logic: being understood by listening.

This involves :

• more natural sentences
• fluid, conversational language
• direct, concise answers

Longer, more involved requests

On average, a voice query contains 4.2 words, compared with 2 to 3 in text. It is more precise, more contextualised and closer to the action. Result: a much stronger intention to buy or act.

Performance, a decisive criterion

Voice responses must be instantaneous. They load on average 52 % faster than traditional web results. Content that is too heavy or poorly structured is penalised. In practical terms:

• choose fast pages
• structure your answers clearly
• avoid unnecessary overload

Vocal GEO is no longer a matter of anticipation; it is already a lever for differentiation. Brands that optimise for these uses today are taking a strategic lead.

Case study: how a brand can benefit from GEO

Let's take a B-to-B software company.

Classic SEO approach :

• optimised pages for key words
• traffic-oriented blog posts

GEO approach :

• creation of complete guides answering business questions
• integration of benchmarks and customer feedback
• structuring into conversational FAQs
• production of sector studies

Expected result:

• greater likelihood of being quoted in AI responses
• strengthening brand credibility
• increased visibility on high-intent queries

The limits of GEO

GEO has aroused a great deal of enthusiasm, some of it excessive. However, there are a number of limitations:

• lack of transparency in AI systems
• instability of practices (rules change quickly)
• risk of dependence on AI platforms
• variable reliability of the content generated

GEO is promising but still immature.

GEO and SEO: towards a hybrid strategy

Should we abandon SEO? Certainly not. As we have seen, GEO is not fundamentally opposed to SEO. Even if AI chatbots don't work like traditional search engines, these two disciplines share a great deal in common.

It's worth remembering that the «death of SEO» has been announced many times since its inception. Yet it is still essential to online visibility, even if it has to adapt to changes on the web. While AI searches are on the rise, traditional search engines are still the main acquisition channel. But ignoring GEO would be a strategic mistake. The right approach is therefore to :

• maintain a solid SEO base
• gradually integrate GEO
• testing formats adapted to AI

What marketers need to do now

To anticipate this change, there are three key actions:

1. Auditing content Are they usable by AI?
2. Rethinking editorial production less volume, more depth
3. Experiment testing formats, monitoring results and making adjustments

So GEO is not the end of SEO, but it does mark a major change. Visibility is no longer just a question of the results pages, but of the responses themselves and, increasingly, of oral, instantaneous, no-click responses. For marketing departments, the challenge goes beyond technology. They need to rethink the way they produce content, demonstrate their expertise and capture attention in an environment where intermediation by AI is becoming central. Brands that know how to adapt early will be ahead of the game.