Google rankings and Bing results are no longer the only places where your business can be recommended. Since 2024, generative engines — Perplexity, ChatGPT Search, Gemini Advanced, Copilot — answer complex questions by citing sources. Those citations work differently from organic rankings. An article ranked first in Google is not guaranteed to appear in an AI response. Conversely, pages absent from the Google top 10 regularly feature among cited sources.
This phenomenon has a name: Generative Engine Optimisation, or GEO. In 2026, ignoring GEO means leaving an entire visibility channel without any management.
Why GEO is not rebranded SEO
Classical search engine optimisation targets a ranking. The goal is to appear in the first results for a given query. A generative engine, on the other hand, composes a response. It selects sources that allow it to formulate a reliable, structured and verifiable answer — not necessarily the best-ranked ones.
Technically, most of these engines use a RAG (Retrieval Augmented Generation) mechanism: they retrieve a set of potentially relevant documents, then a language model synthesises them to produce the response. Selection signals include content structure, data accuracy, density of cited sources, and presence on third-party platforms (comparison sites, specialist forums, sector publications).
A study published in 2024 at the KDD conference by researchers from Princeton and their collaborators measured the impact of different writing techniques on visibility in generated responses. The results show that adding sourced statistics, authoritative citations, and direct answers to specific questions significantly increases the chances that content is picked up by a generative engine. These are precisely the elements missing from the majority of French commercial pages.
An already significant channel
Adoption figures illustrate how rapidly the shift is happening. According to eMarketer (January 2026), 54% of US marketing professionals plan to allocate a dedicated budget to GEO within the next three to six months. This signal does not predict French adoption rates, but it indicates the direction: GEO is moving from a technical curiosity to a budget line item.
On the audience side, Google AI Overviews appeared across a growing share of queries in 2026. This is more pronounced on high-intent informational queries — precisely where your sector expertise has the most value to demonstrate. A training organisation, a software publisher or a consulting firm that does not feature in these panels cedes ground to its competitors.
What this changes in practice
A concrete example. An industrial mid-sized company is looking for a NIS2 compliance provider. They ask Perplexity: "Which providers specialise in NIS2 compliance for French industrial SMEs?" The generative engine does not consult a ranking. It looks for pages that answer that question precisely, with credibility signals: sector references, data backed by France's national cybersecurity agency (ANSSI), FAQs addressing common questions in the domain.
If your service page contains only a list of offerings with no sourced data or sector examples, it will not be cited. A page ranked lower in Google, but written with precision and enriched with verifiable references, is more likely to appear in the response.
Four principles for adapting your content
Answer directly the questions your clients ask. Generative engines favour content that formulates a clear answer in the opening paragraphs. A page that spends five paragraphs on context before reaching the subject will be treated less favourably.
Source your claims. A page that cites ANSSI, ENISA or an official report carries more credibility for a language model than a page asserting figures without attribution. If your service draws on a regulatory framework or standard, cite the text.
Structure with semantic markup. Hierarchical headings, lists, comparison tables and FAQ markup (schema.org FAQPage) ease content extraction. These elements also improve classical search rankings: GEO and SEO are not opposing strategies — they share the same technical foundation.
Build your presence on third-party sources. Publications in sector media, contributions to professional forums, and citations in comparison sites all increase the authority signal. Generative engines cross-reference multiple sources before formulating a response: a brand that appears only on its own website remains poorly visible in AI responses.
EU AI Act Article 50 and citation transparency
A regulatory element reinforces the stakes. Article 50 of the EU AI Act, applicable since August 2026, requires providers of interactive AI systems to identify AI-generated content and make the sources used visible. This transparency obligation makes citation practices more accessible to users, and by extension increases the commercial value of being cited. A source regularly picked up by ChatGPT or Perplexity gains direct visibility with its target audiences.
For a full reading of the transparency obligations governing AI systems in Europe, our article on Article 50 of the EU AI Act and the Digital Omnibus details what remains applicable after the 2026 deferrals.
Where to start
Transitioning to a combined SEO and GEO strategy does not require starting from scratch. It begins with an audit of your existing content: which pages directly answer specific questions? Which contain sourced data? Which have complete structured markup? This audit can be completed in a few weeks and produces a clear prioritisation.
High commercial-potential pages — service pages, case studies, comparison pages — should be addressed first, as they concentrate the purchasing intent that generative engines are beginning to channel.
If you wish to assess your website's visibility level in generated responses and structure an adapted content strategy, our SEO and GEO service covers the initial audit, priority setting, and deployment of optimisations.