Loading page
Five truths about AI search shared at the SEO & GEO Summit in Paris, and the 4-step GEO framework Qwairy gives every client to get recommended by ChatGPT, Gemini and other AI engines.
What actually works in GEO right now? At the SEO & GEO Summit at Disneyland Paris, one of the biggest search events in France, Qwairy shared its answer on stage alongside two clients: a marketing agency and Nissan. This article sums up what those talks covered: five truths about AI search that should shape any GEO strategy, then the four-step framework Qwairy recommends to every client to get recommended by ChatGPT, Gemini, Claude and the other AI engines.
💡 TL;DR
There is no single AI: ChatGPT and Google AI Overviews share only about 7% of their sources. AI answers are volatile, the rules change overnight, conversations go beyond the first prompt, and a mention can still hurt you. A solid GEO strategy follows four steps: choose the prompts your buyers really use, publish the content AI needs, get mentioned on the sources AI trusts, and measure revenue, not just AI traffic.
I gave 3 talks on GEO last week. Here's the best of it.
GEO (Generative Engine Optimization) is the practice of making your brand visible, cited and recommended in the answers of AI engines such as ChatGPT, Gemini, Perplexity, Claude or Google AI Overviews. Before building a GEO strategy, you need to accept how different AI search is from classic search.
In a large-scale analysis run on Qwairy data, ChatGPT and Google AI Overviews shared only 7% of their sources. When both engines search the web and cite pages, more than nine sources in ten differ from one engine to the other. The full study is available here: AI engines citation overlap . This is a major break with SEO. For years, search meant Google, which held around 90% of the search market in some countries. If you performed well on Google, you usually performed well on Bing and the rest. In GEO, that shortcut no longer works. You need to be granular:
Each engine has its own sources, so each one needs its own analysis.
AI answers are volatile. Run a prompt today, run it again tomorrow, and you may get a different answer with different brands and different sources. A study conducted with the PR agency Baloo PR ran the same prompts once a day for 7 days. Result: only about 50% of the brands and sources came back from one run to the next. [À VÉRIFIER: exact wording of the 50% finding in the Baloo PR study] The practical rule:
In August 2026, Reddit's share of ChatGPT citations collapsed by 95% from one day to the next. Reddit then came back, before dropping again. The lesson: you cannot bet everything on one source or one platform, because the engines change their rules without notice. The most resilient approach is to be present on a large number of websites across the web. Authority still matters, and higher-authority sites are most probably better, but breadth of presence is what protects you from sudden shifts.
Most of the time, ChatGPT ends its answer with a question. [À VÉRIFIER: the talk cites 95% of answers, while Qwairy's June 2026 study measured about 67% on commercial questions] Ask for a surfboard and the assistant gives you a first selection, then asks for your budget to refine it. The second answer, the one built after the follow-up, is often where the real recommendation happens. So your prompts should be treated as conversations, not as isolated queries. This is why Qwairy built Follow-Ups, a feature that extracts, for every tracked prompt, the follow-up question each AI engine suggests. A few examples:
Each follow-up is a next prompt your buyers are likely to run, and a prompt worth tracking.
Run a free audit: in a few minutes, ChatGPT, Gemini and Perplexity answer 8 buying questions about your market live, and you see who they recommend.
In the early days, GEO was simple: are you mentioned on a prompt or not? That is no longer enough. If ChatGPT mentions your brand and then explains that it is not a brand people can trust, the mention works against you. Two layers matter beyond the mention:
Example: an AI answer states that the Nike Revolution 7 is priced around $60 to $75. If the actual price were $120, that is a piece of information to correct. In Qwairy, you can open the claim, see where it was mentioned and look at the sources behind it. The mention is only the visible part of the iceberg. What AI says about you matters just as much.
Here is the method Qwairy shared during the summit and recommends to every client.
Start from what you sell, then imagine what your prospects ask an AI when they are in the decision phase. If you sell surfboards, your prompt set could include:
Adding personas and use cases to your prompts makes them closer to real conversations.
Do not forget branded prompts. Once people have picked a brand, many go to ChatGPT to double-check their choice before buying. Track prompts such as:
This is especially true for high-value purchases. Nissan, which shared the stage at the summit, is a good example: when a car costs well over 5,000 to 10,000 euros, buyers almost always check their choice with an AI first. Once your prompts are defined, add them to a GEO tool such as Qwairy to monitor them across engines.
Once your prompts run, look at the sources the LLMs use. They show you which topics and formats you should publish. Two patterns stand out today:
Step 3 is off-site. Look at the sources ChatGPT and the other engines cite on your prompts, then go where they go.
Content Opportunities lists the topics where AI mentions your competitors and not your brand, each with an impact score.
LLMs send very little traffic compared to classic search. Judging GEO on clicks alone will undervalue it. That is why Qwairy connects to analytics tools such as Google Analytics, Piwik, Matomo and Piano, as well as Google Search Console, to show:
A buyer can discover you in ChatGPT, then search your name on Google and buy from there, or buy from another website. Your analytics will credit Google, even though the discovery happened in an AI conversation. The simple fix: whenever someone buys on your site, ask how they discovered you, with a short form after purchase. It is the most direct way to capture AI influence that attribution tools cannot see.
Ready to build a GEO strategy that drives results? Try Qwairy for free and get 120 credits to track and improve your visibility across ChatGPT, Gemini, Perplexity and other LLMs.
Track the questions that matter to your business, see which brands each answer mentions and which sources it cites, then work on the gaps.
Free plan with 120 initial credits • No credit card required
Other Articles
The GEO Mistake Everyone Makes: Tracking Only the First AI Answer
Most AI visibility checks stop at the first answer. But AI assistants turn questions into conversations, and the follow-up questions they suggest decide which brands stay in the race.
Answer Engine Optimization (AEO): The Complete 2026 Guide
A complete, practitioner-level guide to Answer Engine Optimization: what AEO is, how answer engines like ChatGPT, Perplexity and Google AI Overviews select and cite content, the core levers, a step-by-step workflow, and how to measure it.