In short: GEO — generative engine optimization — means setting up your business so that AI assistants like ChatGPT, Gemini, Perplexity and Google's AI Overviews mention it and cite it as a source. This is no longer a niche: according to BrightLocal's Local Consumer Review Survey 2026 (US), 45 percent of respondents now get local recommendations from AI tools — a year earlier it was 6 percent. GEO builds on SEO but weighs things differently: only 38 percent of pages cited in Google's AI Overviews also rank in the top 10 for the same query (Ahrefs). For restaurants, hotels and local businesses, four things matter most: reviews on more than one platform, a website that serves as a clear source of facts, mentions by third parties, and content with verifiable numbers. What doesn't help: an llms.txt file as a silver bullet — Google doesn't use it.
What is GEO — and how is it different from SEO?
The term comes from a study by researchers at Princeton University and other institutions, presented at the KDD conference in 2024. It covers every measure that increases a source's visibility in AI-generated answers. You will also see the terms AI SEO, LLM SEO or AEO (answer engine optimization) — they mean the same thing.
| SEO | GEO | |
|---|---|---|
| Goal | a position in the results list | a mention or citation in the AI answer |
| Unit | the page | the statement, the fact, the business |
| Key signals | relevance, links, technology | clear facts, agreement across sources, reviews, freshness |
| Measurement | rankings, clicks | mentions in test prompts, AI report in Search Console, visits from chatgpt.com |
Important: GEO does not replace SEO. 38 percent of cited pages rank in the top 10 — so a good ranking remains an advantage. But almost two thirds of citations come from elsewhere, a large share even from pages that don't rank within the first 100 results for that query. If you only look at rankings, you no longer see a growing part of your visibility.
Where AI gets what it says about your business
Google (AI Overviews, AI Mode, Gemini) draws on its own search index and on data from Google Maps and the Google Business Profile — including your reviews, opening hours and photos.
ChatGPT, Perplexity and others search the open web. For local queries, a BrightLocal analysis found that 58 percent of ChatGPT Search's sources were business websites, 27 percent mentions on other sites and 15 percent directories. BrightLocal also points out that AI tools like ChatGPT cannot see inside Google's walled garden of reviews — which makes reviews on other platforms such as TripAdvisor more important.
And they recommend well-rated businesses. According to SOCi's Local Visibility Index 2026, the local businesses AI assistants recommend average 4.3 stars on ChatGPT, 4.1 on Perplexity and 3.9 on Gemini.
The consequence: an AI recommends what several sources agree on. A single perfect signal is not enough — it takes a consistent picture across many sources.
The seven levers for restaurants, hotels and local businesses
1. Reviews — on more than one platform
Many fresh, answered reviews are among the strongest local signals you can influence for Google — and for AI assistants the most important evidence that a business is good. Besides Google, maintain at least one second platform that matters in your industry: TripAdvisor for restaurants and tourism, Booking.com or HolidayCheck for hotels. Reply specifically — a reply that picks up the dish or the occasion adds context. How to do this systematically is covered in the complete review guide.
2. Your website as a source of facts
AI systems cite sentences that answer a question. Nothing can be cited from "Passion for fine food since 1998". From this sentence, plenty: "Italian restaurant in Lucerne's old town, 60 seats, terrace on the Reuss, mains CHF 28 to 42, gluten-free pasta on request, Monday to Saturday 11.30 am to 11 pm." Put a fact paragraph like this on your homepage, the menu as text on the website and the key details as structured data in the code. What else a restaurant website needs is covered in Having a restaurant website built.
3. Answer blocks and verifiable numbers
The Princeton study tested which text changes increase visibility in AI answers. Adding statistics, quotations and source citations worked best — with up to 40 percent more visibility. Classic keyword stuffing, by contrast, did nothing. For your website that means: questions as subheadings, the answer in the first sentence, numbers with source and date. A FAQ block with your guests' real questions is the simplest format for this.
4. Mentions by third parties
Local press, restaurant guides such as Gault&Millau, the Michelin Guide or Falstaff, tourism organisations, city blogs, lists like "The best brunch spots in Zurich": every credible mention is one more piece of evidence. You can't buy it — but you can trigger it: with a real occasion (new seasonal menu, anniversary, award), a surprising number from your business or an up-to-date listing with the local tourist office.
5. A consistent identity
Name, address, phone, category and short description must read the same everywhere: Google, website, Local.ch, search.ch, TripAdvisor, booking platforms, social media. If sources contradict each other, the AI becomes uncertain — and when in doubt names another business. Details in Improving your Google ranking.
6. Video
YouTube is one of the most cited sources in AI answers. According to Ahrefs, 18 percent of AI Overview citations that don't come from the regular top results come from YouTube. Short videos of your kitchen, signature dish, rooms or how to get there — with a descriptive title and text — are an underrated lever.
7. Technology: don't lock out AI crawlers
Check your robots.txt: if you block "OAI-SearchBot" (ChatGPT search) or "PerplexityBot", you can't appear there as a source. Google's AI Overviews use the regular Googlebot. Some hosting and CDN providers now block AI crawlers by default — a quick look at the settings is worth it.
What doesn't work
llms.txt as a silver bullet. The file is meant to give AI systems an overview of a website. Google, however, has stated that it uses it neither for search nor for AI Overviews. We have one ourselves — it does no harm, but don't expect an effect.
Keyword stuffing and mass-produced AI texts. In the Princeton study, keyword stuffing brought no improvement. A hundred generic blog posts without facts of their own offer nothing an AI would need to cite.
Bought reviews. They distort exactly the signal AI assistants rely on — and Google deletes them. The risks in detail.
One-off pushes. AI systems prefer current information. One fact update a year is too little; reviews, opening hours and offers have to be right continuously.
How to measure whether GEO works
1. The monthly prompt test. Define ten fixed questions the way your guests would ask them — for example "Where can I have brunch with a lake view in Lucerne?" or "Good seminar hotel on Lake Zurich for 20 people?". Ask them once a month in ChatGPT, Gemini, Perplexity and Google's AI Mode, in a private browser window. Note: mentioned yes or no, in which position, which source was cited.
2. Search Console. Google's Search Console now links to a dedicated report on performance in generative AI features.
3. Your web analytics. Watch for visits from chatgpt.com, perplexity.ai and gemini.google.com. ChatGPT often appends the parameter utm_source=chatgpt.com to links — which makes these visits easy to count.
The 30-day plan
Week 1 — Inventory. Define the ten test questions and run them a first time. Check robots.txt and CDN settings. Align name, address and phone on Google, website, Local.ch and search.ch.
Week 2 — Facts on the website. Fact paragraph on the homepage, menu as text, a FAQ with eight real guest questions, structured data for restaurant or hotel.
Week 3 — Reviews. Reply to every review of the last 90 days, on Google and on a second platform. From now on: reply to new reviews within three days.
Week 4 — Third parties and video. Create three occasions for mentions (update your tourist office listing, a real story for the local paper, maintain your restaurant guide profile) and publish a first short video.
After that: the prompt test once a month — and the results logged in a simple table.
What we did on our own website
We implemented these levers on our own site first: our guides open with an answer block and close with FAQ and sources, the FAQs are marked up as structured data, and AI agents can fetch every page as a lean text version. Since mid-September 2026 our Search Console has shown clearly more impressions. How much of that is due to GEO and how much to new content can't be cleanly separated — we would rather say so openly than claim an effect we can't prove.
FAQ
What is generative engine optimization (GEO)? GEO stands for generative engine optimization: measures that make AI systems like ChatGPT, Gemini, Perplexity or Google's AI Overviews mention your business in answers and cite it as a source. The term comes from a Princeton University study presented at the KDD conference in 2024.
Does GEO replace classic search engine optimization? No. GEO builds on SEO — a good ranking increases the chance of being cited. But only 38 percent of pages cited in AI Overviews rank in the top 10 for the same query. Reviews, third-party mentions and clear facts are gaining weight.
How does ChatGPT come to recommend my restaurant? ChatGPT recommends what several sources agree on: your website with clear facts, review platforms, restaurant guides, local press. Because, according to BrightLocal, AI tools like ChatGPT can't see directly into Google's reviews, reviews on other platforms such as TripAdvisor matter especially.
Do I need an llms.txt file? It does no harm, but as things stand it brings little. Google has stated that it uses the file neither for search nor for AI Overviews.
How do I measure my visibility in AI answers? With a monthly prompt test (ten fixed questions, several AI tools, logged), the report on generative AI features in Google Search Console, and visits from chatgpt.com or perplexity.ai in your web analytics.
What does GEO cost? The basics — facts on the website, consistent listings, replying to reviews — mainly cost time. The biggest lever for local businesses is a steady stream of real, answered reviews; you can organise that yourself or hand it to a system.
Sources
- Aggarwal et al.: GEO: Generative Engine Optimization, KDD 2024 (arxiv.org/abs/2311.09735)
- Ahrefs: Update: 38% of AI Overview Citations Pull From The Top 10 (ahrefs.com/blog)
- BrightLocal: Local Consumer Review Survey 2026 (brightlocal.com/research)
- BrightLocal: Uncovering ChatGPT Search Sources (brightlocal.com/research)
- SOCi: Local Visibility Index 2026 (soci.ai)
- Google Search Central: AI features and your website (developers.google.com/search)
- Search Engine Roundtable: Google Search Team Does Not Endorse LLMs.txt Files (seroundtable.com)
Related reading
- AI Overviews are quietly erasing your bookings — what Swiss hoteliers must do
- Improving your Google ranking: local SEO for restaurants and hotels
- Having a restaurant website built: costs and checklist
- Google Ads costs in Switzerland
The strongest GEO lever for restaurants and hotels is a steady stream of real, answered reviews. Trophy gives every server a personal NFC card and answers reviews automatically in Swiss tonality — a 10-minute demo, your venue as the example.