With the rise of generative AI and the advent of GEO (Generative Engine Optimization), the online search landscape is becoming increasingly complex. SEO, GEO and affiliate partnerships can no longer be thought of in silos: it’s their articulation that will make the difference.
Together, these areas of expertise build a coherent ecosystem, capable of ensuring immediate performance in Google, while preparing for tomorrow’s presence in generative environments.
Part 1: SEO, GEO: AI commerce reshuffles the deck
Before getting to the heart of the matter, can you define the main terms SEO, SGE, GEO?
Sébastien Barry: Yes, it’s true, we use a lot of acronyms to talk about our business. SEO (Search Engine Optimization) is all about understanding how the algorithms of “classic” search engines like Google or Bing work, and implementing strategies to gain visibility. It’s a discipline that’s been around for some twenty years, and is now familiar to almost everyone.
What’s new is SGE (Search Generative Experience), which Google launched with the rise of artificial intelligence. In concrete terms, this means integrating AI-generated answers directly into traditional search engine results. The challenge for brands is to successfully appear in these new AI-generated inserts: to highlight their products, services and brand.
Finally, there’s GEO (Generative Engine Optimization), which is a kind of “new SEO”, but designed for AI engines based on LLMs (Large Language Models). Here, the question is no longer simply “how to rank well in Google”, but “how to be cited, visible and credible in the responses generated by these new AI engines”.
We read here and there that “SEO is dead”. Has GEO already replaced SEO?
Sébastien Barry: If we take a step back, it’s worth remembering that SEO has been regularly heralded as “dead” for over ten years. And yet, it’s constantly evolving, and today remains indispensable. Will it disappear this time? It’s hard to say. But one thing is certain: Google remains a central player for advertisers. No one imagines stopping using Google as an acquisition channel tomorrow. They have mastered AI, know how to integrate it into their products, and have considerable financial resources at their disposal. It seems naive to think that Google will let itself be overtaken in this field.
At the same time, new engines and new uses are emerging, driven by LLMs. These platforms, still marginal in terms of traffic, are experiencing dazzling growth, far more rapid than what we experienced when the Internet first arrived. Adoption speeds are impressive, and it’s essential to start taking an interest in them, understanding them and experimenting, as these are ecosystems destined to play a major role.
That said, Google is still growing: search volume continues to increase despite the rise of LLMs. What we’re seeing is not so much a complete transfer of uses as a diversification of paths and entry points. Users are testing out different tools, but Google remains central. It should also be remembered that the adoption of generative AI is still very relative: in France, some 40-45% of Internet users have never used ChatGPT or similar services.
The real question, then, is not whether SEO is going to die, but rather: how can we influence both traditional engines and LLMs?
On this point, the levers remain largely similar between SEO and GEO (Generative Engine Optimization). Technical and semantic aspects still count, even if authority plays a slightly different role. But overall, it’s difficult to perform well in LLM without already mastering SEO. GEO doesn’t replace SEO: it adds a layer of complexity, new adjustments, but rests on the same solid foundation. So I’d say that SEO isn’t dying, it’s changing, as it always has. We’re almost back to the original spirit: confronting an algorithm, trying to understand it, deconstructing it and extracting performance from it. Whether it’s Google, another engine or an LLM, the approach remains the same. SEO professionals who maintain this inquisitive, test-and-learn approach, and this ability to adapt, will continue to perform flawlessly in this new ecosystem.
What is the operational definition of GEO? What does it mean to be cited or sourced by an AI engine? Are there any new metrics to track?
Thibaut Fitoussi: GEO (Generative Engine Optimization) is the global optimization of a site, its structure and content, but also of a brand’s entire digital ecosystem.
The aim is to work on several aspects. First, technical optimization. LLMs particularly appreciate structured, standardized data. For example, the integration of schema tags is essential. In fact, some of these best practices are common to classic SEO. Next, semantic optimization. This is where GEO platforms play a key role. They help identify “visibility gaps”, i.e. topics or queries on which a brand is absent, while its competitors are present. These gaps become opportunities for positioning. Finally, offsite optimization. Unlike traditional SEO, which focuses on backlinks, GEO is based on the publication of content by external publishers, whose information naturally appears in LLM responses. The logic is therefore no longer purely quantitative (number of links), but qualitative: being cited by the right sources.
As far as GEO metrics are concerned, as the market is still new, there is no universal standard. Each platform develops its own KPIs, but some common benchmarks are emerging:
- Mention rate: this is the central indicator, measuring the number of times a brand is cited in the responses generated by LLMs. It’s a new-generation form of ranking.
- Position in responses: like traditional SEO ranking, we analyze where the brand appears in the generative engine response.
- Missed hits: we measure the number of relevant queries where the brand does not appear. The aim of GEO is precisely to reduce this rate over time.
- Qualitative indicators: depending on the platform, we also find sentiment metrics (positive/negative/neutral), or other more specific indicators.
In short, GEO takes up some of the fundamentals of SEO, while integrating new levers and indicators adapted to AI engines and LLMs.
Traffic generated by AI engines is still low. Should we wait before taking action dedicated to GEO and, for brands, is it risky to wait?
Christophe Bosquet: I don’t think there’s any real risk in waiting too long. In fact, the projects we need to launch today are often directly linked to SEO, so they’re still useful whatever happens in the market.
In concrete terms, GEO analyses already highlight needs that also serve SEO: structural site optimizations, editorial and semantic enrichments, and improvements to online brand experience and credibility. All of this reinforces SEO, so it’s never a waste of time.
The right approach is to divide projects into two parts. Firstly, what is necessary today and forms a common base (technical, content, structured data, consumer reviews, updating a Wikipedia entry, etc.), as this benefits SEO immediately.
Secondly, what is more probabilistic, i.e. actions that prepare for the future, on the assumption that they will weigh tomorrow on engines like ChatGPT or other LLMs.
The problem is that the ecosystem is constantly moving. One announcement can be contradicted by another the next day, and nobody can say today what OpenAI, Google or others will be doing in six months’ time. That’s why we need to move forward step by step, securing the present with solid SEO actions, and preparing for tomorrow by testing, implementing GEO tools and prioritizing actions on relevant semantic fields (legitimate vs. less legitimate territories). In short, doing nothing is risky. But moving forward methodically, building a robust SEO foundation and gradually preparing for GEO, means you’ll be ready for both today and tomorrow.
Part 2: SEO, GEO, affiliate partnerships: actions to take
Is there anything you can do today to extend your website’s semantic coverage, reinforce its authority, structure your site so that it’s “AI compatible”?
Sébastien Barry: After the technical basics, the semantic dimension is what really differentiates SEO from GEO, with a nuance on the “authority” part, which remains important but is handled differently. In SEO, the objective is clear: identify the transactional queries that generate the most traffic and value, then build a site structure and content capable of performing on these strategic keywords. The logic is therefore to focus on high-volume priority queries.
In GEO, the logic changes: when a user types a prompt, the LLM actually executes several searches in parallel on similar semantic fields. Here, it’s no longer a question of being “the best on a single query”, but of being present on as many sub-queries as possible in order to obtain the best weighted average visibility. In other words, those who cover the widest semantic territory will be cited. This is what makes GEO particularly interesting, as it forces us to think about our territories of legitimacy. What are the themes where my brand should appear, but where I’m absent and where my competitors are taking over?
This transforms your content strategy. You need to cover all the questions your persona might have, not just the high-volume keywords. Content must be clearer, more direct and structured than traditional SEO articles, which are often long and complex. The use of headings (Hn), bulleted lists, tables and structured data is highly recommended, as LLMs extract information very quickly from these formats. LLMs are also very sensitive to fresh, unpublished and quantified data: recent statistics, market studies, dated data, but also clear attribution of the author to reinforce trust. GEO performance therefore requires broad, structured and up-to-date semantic coverage, rather than focusing on a few strategic queries.
And for brands, what actions should be carried out off-site, within their ecosystem?
Thibaut Fitoussi: We’re no longer in a situation where the main objective is to win backlinks, as it used to be. Of course, this can still be integrated into an overall strategy and unified with other efforts, but the logic has changed. Today, sources are much more varied. On our platform, we see hundreds of different sources. The challenge is no longer simply to obtain links, but to ensure a presence in the corpora that AIs consult and cite. In the feedback we’re seeing, there’s still a fairly classic component, close to traditional SEO, with quality, high-authority publishers. But a new ecosystem is gaining in importance: that of UGC sites. This is a major evolution that is overturning traditional approaches. These include conversational spaces such as Reddit, but also GitHub, specialized forums, review sites, as well as document databases and product data. Anything that gives the LLM access to structured information is particularly valued. The more enriched and well-organized the content, the more likely it is to be exploited by these models.
These sources must be crawlable by LLMs, which presupposes that they are accessible, that they do not block their indexing via Cloudflare or robots.txt, and that they offer a clear structure. Increasingly, it’s the structuring of information that determines the quality of content, far more than its sheer quantity. The idea, as in SEO, is not to multiply points of presence, but to focus on relevance and quality. This is precisely what we do with the platform: identify the most strategic sources, those that correspond to a brand’s DNA and enable it to express itself in a favorable environment. It’s no longer just a question of being present with the big publishers of yesteryear, but of building a balanced mix between these authoritative references and conversational spaces like Reddit, where the corpora exploited by AIs are increasingly being formed.
How to activate useful GEO partnerships? How can you orchestrate publishers, comparators, creators, communities, etc. to multiply signals to AI engines?
Christophe Bosquet: We’re gradually moving away from the classic logic of partnerships as purely audience leverage, to one where partnerships also become a means of obtaining semantic presence among sources exploited by LLM engines. In the DIY sector, for example, we analyzed the best French e-commerce sites. One of the market’s major players didn’t appear in the results. Digging deeper, we discovered that ChatGPT relied on an obscure shopping guide, barely 100 visits per month according to Similarweb, non-existent in Google results, but recognized by LLM as a legitimate source. As this player was not listed, it was simply absent from the answers. This is where the work begins: differentiating partners according to whether they have a significant audience, in which case affiliation can be a lever of activation, or whether they don’t, but offer semantic legitimacy. In the latter case, the aim is not to obtain direct traffic, but to ensure a presence in the sources that search engines consider relevant (editor’s note: read our article: The hidden value of niche affiliates in the age of AI). This can take the form of editorial partnerships, conversations on Reddit, Substack or other community platforms, where the value lies not in the link but in the presence at the heart of exchanges. Our role is therefore to identify the types of sources that are missing – buying guides, community platforms, high-traffic sites – and then activate the right partnerships according to the context. Sometimes it’s affiliation, sometimes it’s SEO-friendly backlinks, sometimes it’s purely editorial content designed to feed the search engines. The challenge is to build an intelligent network that combines economic efficiency with editorial relevance.
Part 3: SEO, GEO, affiliation: how can they work together?
How can an SEO agency leverage AI to generate large-scale content while maintaining a unique voice, which is favored by search engines? How do you juggle the mass of content you can produce with the quality you need to ensure?
Sébastien Barry: It’s always a delicate balance. AI is extremely interesting in its ability to accelerate production processes, whether in data analysis or content creation. But it’s important to remember that AI is only a tool: it’s not a strategy. Strategic thinking needs to take place upstream, ideally with the support of a partner capable of defining the right objectives, identifying the content to produce, the actions to implement and the direction to follow. It’s only then that AI comes into its own, streamlining processes, accelerating production and saving time. AI becomes a formidable tool as soon as it is properly framed from the outset, and every output is reread, checked and validated before going online. On the other hand, when this framework is lacking, drifts quickly appear. We’ve already seen cases where a button was simply pressed to generate a mass of content, which on the surface seemed useful but ended up damaging the site. In the absence of clear direction, quality declines, duplicate content multiplies, content cannibalizes itself and performance plummets. AI must therefore be seen above all as a performance lever, integrated into an overall strategic reflection, and not as a simple shortcut to produce quickly, at the risk of producing poorly.
How does a GEO platform help marketing teams better predict user needs and adapt their semantic content in real time?
Thibaut Fitoussi: In reality, we’re doing the opposite of LLM. When a user asks a question, an LLM will rely on a set of sources to formulate its answer. However, these models are insensitive to unfounded marketing arguments: a brand can repeat that it is the best, but this has no impact. What counts are factual, tangible arguments found in content published on different platforms, sometimes reliable, sometimes less so. Our approach is to go full circle: we identify the arguments exploited by the LLM, then build counter-arguments to fill in the gaps in the brand discourse. The idea is to help the company take up a position on issues where its competitors are better represented, and to enrich its narrative with solid evidence. This is exactly the role of a GEO platform: to act as a semantic catalyst. The platform generates optimized, structured and “IA friendly” content, with verifiable data and a log of evidence, integrating expert quotes, producing comparative tables aligning arguments against those of competitors, respecting best practices identified by benchmark studies (Princeton, etc.).
The aim is twofold: to produce high-performance content for the brand and to respond to the formats favored by LLMs. This is because, depending on the subject, the models rely on different types of content such as listicles (practical lists), reviews and user opinions, news and current affairs, or tutorials and practical guides. The GEO platform analyzes these uses and guides the brand towards the right content typology, reversing the logic: starting from what the LLM is already using, to reinforce the brand’s presence and legitimacy in the face of its competitors.
How do you get this trio – SEO agency, GEO agency and partnership agency – to work together?
Christophe Bosquet: Historically, SEO on the one hand, and partnerships and affiliation on the other, were seen as two parallel universes, with few bridges between them. With the arrival of GEO, this separation has disappeared: SEO can no longer work alone. It must rely on sources, partners, and a more global approach where the different levers reinforce each other. In the digital world, everything is interconnected. An influencer campaign, for example, can be highlighted in Google Discover, feed LLM engines, and so on. Each action feeds both traditional organic results (like Google) and new exposure environments.
This complementarity is key: when missing sources are identified on the GEO platform, we can decide to activate them via SEO (backlinks and visibility logic) or via performance and affiliation (audience and conversion logic). In all cases, teams need to work together. We can no longer operate in silos, with an SEO team on one side, an affiliation/partnership team on the other, and an isolated SEA team. Everything is connected. Direct and indirect organics are taking on new importance: content produced in one place will be exploited elsewhere. A striking example of this is Google Discover, which features posts by influencers: the social network becomes a resource for Google, proving that everything is intertwined.
In short, the complementary nature of SEO, affiliation, partnerships, SEA and influence is no longer an option: it’s a given. The challenge is to stop working in silos and to think of each piece of content as a multi-purpose lever, capable of feeding several channels simultaneously.
Do you have a priority action you’d like to recommend to start building a presence on AI engines?
Sébastien Barry : To optimize a product or service page, an effective approach is to cross-reference it with your persona using an LLM. The idea is to practice reverse engineering: you submit the page to the engine, specifying the persona you’re targeting, and ask it what questions this persona might be asking, and to which the page already provides an answer. The result is a series of relevant questions. These questions can then be fed back into the LLM, but this time in the other direction: you want to know what keywords a user would type into Google to obtain these answers. You then use these keywords in Google, analyze the results and retrieve the “People Also Ask” to further expand the range of possible queries.
By comparing all this exploratory work with your existing content, you can clearly see where you’re already well positioned, where you’re lacking, and what new content needs to be produced. This method transforms a simple page into a real lever of visibility, consistent with the expectations of your personas and aligned with the uses of both traditional and generative search engines.
Thibaut Fitoussi: With the brands we work with, we always observe cycles. Today, digital is undergoing a new shift, a new transformation that is giving rise to a great deal of reflection and generating a heavy mental load for marketing teams. They’re thinking that they’re going to have to develop additional expertise, recruit new profiles, invest further in an area whose effectiveness has yet to be proven, either in terms of traffic or return on investment. This is precisely what makes this stage so difficult to grasp. The first recommendation is to produce content that is truly Geo Friendly, i.e. that respects best practices and fills the semantic gaps left by the brand, while remaining at the service of SEO. The aim is to unify SEO and GEO efforts so that, even if ROI is not immediate, each action contributes to the brand’s overall strategy and avoids dispersion of resources.
Christophe Bosquet: To complete the picture, I’d like to go in the same direction: when it comes to managing partnerships, whether affiliation, influence or collaborations in the broadest sense, the key is to bring the teams together. SEO managers and partnership managers need to work together, with a common logic. In the same way that GEO feeds SEO and SEO feeds GEO, it’s essential to move away from a siloed approach. The first concrete action to take, starting tomorrow, is therefore simple: organize regular meetings between the SEO and partnerships teams to build shared strategies rather than working in parallel.
