We use analytics to improve the site. Privacy
How to optimise your website for AI answer engines (ChatGPT, Perplexity, Google AI Overviews). A GEO guide for Swiss SMEs with checklist, concrete levers, and measurement methods.
In 2026, nearly one in two Google searches ends without a single click to an external site. AI Overviews, enriched featured snippets, and answers generated by AI assistants capture intent before the visitor even reaches your page.
60%
of Google searches are now zero-click searches, up from 50% in 2023
SparkToro / Datos, 2025
For a Swiss SME, this changes the rules. Ranking first on Google no longer guarantees traffic. What matters now is being cited in answers generated by ChatGPT, Perplexity, Claude, or Google AI Overviews. This is called GEO: Generative Engine Optimization.
GEO doesn't replace SEO. It complements it. Both disciplines rest on the same technical foundations (quality content, clear structure, domain authority), but GEO adds a layer of optimisation tailored to how AI models select and cite their sources.
Both approaches share a common goal: making your content visible. But their mechanics differ on several key points.
| Critère | Classic SEO | GEO |
|---|---|---|
| Target | Ranking algorithms | Generative models (LLMs) |
| Success metric | SERP position | Citation in AI answers |
| Content format | Long keyword-optimised pages | Direct answers, sourced data |
| Authority signal | Backlinks, domain age | External citations, brand mentions |
| Technical structure | XML sitemap, meta tags | Schema.org, FAQ, llms.txt |
| Measurement | Search Console, Ahrefs | Profound, Otterly, manual audits |
Concretely, an article ranking first on Google may never be cited by ChatGPT. Conversely, a page barely visible in the SERPs may be systematically cited by Perplexity if it contains a sourced data point that nobody else offers.
Based on what actually works in 2026 to surface a website in AI answers, here are the five levers to prioritise.
Semantic structure as direct answers
Each section must answer a precise question in its first sentence. LLMs prioritise paragraphs that state the answer before the nuances. Ideal format: question as H2, answer in 2-3 sentences, then development.
Sourced quantitative data
AI models prioritise pages that bring numbers with an identifiable source. A statistic without a source won't be picked up. With a source and a date, it has a much better chance of appearing in a Perplexity answer.
Schema.org FAQ and HowTo (when relevant)
FAQPage markup is one of the most consumed by Google AI Overviews. Every major article should have a structured FAQ section with Schema.org. It's also what feeds enriched featured snippets.
llms.txt file at the root
Inspired by robots.txt, llms.txt has become a de facto standard for exposing your site's priority pages, descriptions, and internal links to LLMs. It facilitates indexing by AI crawlers from Anthropic, OpenAI, and Perplexity.
External citations and mentions
LLMs judge a site's authority by brand mentions on other trusted domains, not just by backlinks. A guest post without a link on a recognised media counts as much as a dofollow backlink. Think press relations, podcasts, interviews.
Before launching a content overhaul, run this quick audit on your key pages. If you check fewer than 6 boxes out of 10, your site isn't ready for AI engines.
Express GEO audit
Test your AI visibility in 5 minutes: ask ChatGPT and Perplexity a typical industry question (for example, which web agency in Lausanne for an SME). If your brand doesn't show up, you have a GEO gap to close.
The big GEO challenge in 2026 is measurement. Google doesn't provide a Search Console for AI Overviews, and LLMs don't expose analytics. Fortunately, several tools are emerging.
| Critère | Tool | Use case |
|---|---|---|
| Profound | LLM citation tracking | Paid |
| Otterly.ai | AI Overviews + ChatGPT monitoring | Paid |
| AthenaHQ | Automated GEO audit | Paid |
| Manual audit | Prompt tests on ChatGPT/Perplexity | Free |
For a Swiss SME on a tight budget, the manual audit remains the most accessible method. List 20 questions your prospects ask (for example, how much does a website cost in Lausanne), test them monthly on ChatGPT, Perplexity, Google AI Overviews. Note whether you are cited, which competitor replaces you, and which sources the models favour.
The AI-generated content trap
Many SMEs publish content produced by ChatGPT to save time. AI models detect and downrank generic content with no added value. GEO rewards real expertise, not volume.
GEO isn't a revolution, it's a natural evolution of SEO. The fundamentals remain the same: useful content, clear structure, authority built over time.
What changes is the priority given to clarity and verifiability. A page that directly answers a question with sourced data and clean markup will be cited by LLMs. A page that drowns the information in 3,000 words of filler won't.
Start with your 5 most strategic pages (homepage, main services, FAQ). Optimise them for GEO before touching the rest. These are the pages where your prospects will decide, and they're the ones LLMs will cite first.
SEO and GEO audit of your site
We look at your site through an AI engine's eyes. You leave with a 90-day prioritised action plan.
Request an auditNo. GEO complements SEO. Both rest on the same technical foundations (quality content, structure, authority). GEO adds optimisation tailored to AI models. An effective 2026 strategy combines both.
AI models reindex their sources every 1 to 3 months depending on the tool. Expect 2 to 4 months to see the first effects of GEO optimisation. It's faster than classic SEO but more volatile: a model update can disrupt citations.
Yes, it has become a standard. The llms.txt file exposes priority pages of your site to AI crawlers with a short description. It's the modern equivalent of the XML sitemap for the LLM ecosystem. Easy to implement: 30 minutes.
Yes, and it's even an advantage. Most French-speaking Swiss SMEs haven't yet optimised for GEO. The French-speaking market is less competitive than English, making gains faster to obtain.
Manual method: ask 20 industry questions to ChatGPT and Perplexity each month. Tooled method: Profound, Otterly.ai or AthenaHQ automatically track brand citations in AI answers. Budget: 50 to 200 CHF per month for the tools.
Particularly so. When a prospect asks ChatGPT for a web agency in Lausanne or a hairdresser in Vevey, the model proposes 3 to 5 names. Being on this list is worth far more than position 8 on Google. Local GEO is a major competitive advantage in 2026.
Go deeper on search visibility
Explore our SEO approach to structure content strategy, internal linking and technical optimizations.
View our SEO pageEvery collaboration starts with a conversation. Tell us about your project, we'll get back to you within 48h.
