GEO for Restaurants and Cafes: How to Get Recommended by AI

Anton Tysiachnik, founder of InSync AgencyUpdated: 30 September 20267 min read

An AI assistant recommends a restaurant when it can quickly confirm the basics: where it is, when it's open, what's on the menu, what dinner costs and whether you can book a table. To make that possible, a restaurant needs a text menu on its website, a page with address and hours, a complete Google Maps profile with reviews, and the same contact details on every platform. Restaurant and Menu markup helps machines read this data without guesswork.

What do diners ask AI assistants about restaurants?

People ask ChatGPT and Gemini the way they'd ask a friend: “where to go for a date downtown”, “breakfast near me before 9 am”, “restaurant with a kids' play area”, “Georgian food with outdoor seating”. The assistant pulls options from websites, maps, reviews and local roundups, and names the places that spell out the right details.

Occasion: a date, a birthday, a business lunch, a group dinner.
Cuisine and dishes: “where to get good ramen”, “vegan menu”, “gluten-free dishes”.
Time and place: “open now”, “weekend brunch”, “near the train station”.
Amenities: patio, parking, dog-friendly, kids' menu, live music.
Budget: “cheap lunch”, “average price per person”.

If your site never mentions a patio or breakfast hours, the model has nothing to connect you to those questions. Even if your patio is the best in the neighborhood.

Why should your menu be text, not an image or PDF?

Many restaurants post their menu as a photo or a PDF. That's familiar to people but harder for crawlers: they may not read text inside an image at all, and they don't always parse PDFs correctly. As a result, the AI model doesn't know you serve breakfast or vegan dishes.

Publish the menu as a regular HTML page: sections, dish names, a short description, price.
Label key features: vegetarian, gluten-free, spicy, kids' portion.
Give breakfast, business lunch, the bar and the seasonal menu their own sections.
Keep prices current and show a last-updated date.

You can keep the PDF as a printable version. The page should be the main source for search engines and AI assistants.

Which pages and details does a restaurant website need?

One beautiful homepage isn't enough. Each type of guest question is best answered by its own page or block.

Page	What goes on it	Which questions it covers
Menu	Dishes, prices, dietary labels	“Where can I get…”, “how much is…”
Location and hours	Address, landmarks, parking, hours by day, kitchen closing time	“Near me”, “open now”
Reservations	Form or booking link, phone, booking and deposit rules	“Book a table for…”
Private events	Rooms, capacity, event menu with starting price, terms	“Private dining for…”, “where to celebrate a birthday”
Chef and concept	Who the chef is, the cuisine, where ingredients come from	“Chef-driven restaurant”, “restaurant by chef…”
FAQ	Kids, pets, delivery, deposit, dress code	Follow-up questions

State the average check as a range, for example “… to … per person”. An AI assistant can repeat a line like that almost word for word, and guests immediately see whether the place fits their budget.

Want to know where your website stands today? Free 60-second AI visibility audit in Telegram.

Check my site

How do Google Maps and reviews affect AI recommendations?

For “near me” and “open now” questions, Gemini and AI Overviews rely heavily on Google data, while ChatGPT with search partly draws on the Bing index. Your Google Business Profile becomes one of the main sources of facts about you, so fill it out completely and check the same details in Bing Places.

An accurate category: “Restaurant”, “Cafe”, “Coffee shop”, plus cuisine type.
Opening hours, including holidays, and kitchen hours.
The menu or a link to your menu page, plus a booking link.
Attributes: outdoor seating, delivery, high chairs, card payments, wheelchair access.
Fresh photos of dishes, the dining room and the entrance.

In reviews, guests name dishes and occasions themselves: “best breakfast in the area”, “celebrated our anniversary here”. Those words can help link your place to real questions. Ask for reviews naturally, with a QR code on the check or a word from the server after dinner. Reply to every review, criticism included, calmly and to the point.

Why must your contact details match on every platform?

Your restaurant shows up in many places: your site, Google Maps, Instagram, delivery and booking apps, city guides. If one lists old hours and another an old phone number, the AI model may give a guest wrong information or fail to recognize that it's the same place. Your name, address and phone (NAP) should match word for word.

Pick one spelling of your name, including a Latin-script version if you need one.
Check address, phone and hours on your site, maps, social profiles and delivery apps.
Changing hours for a season or holiday? Update every platform on the same day.
For a chain, create a separate page for each location.

If AI keeps recommending the place next door, see why ChatGPT recommends your competitors instead of you. For restaurants abroad where guests search in more than one language, read GEO for businesses abroad.

Which Schema.org markup does a restaurant need?

JSON-LD markup hands machines your key details in a structured form. These types fit a restaurant:

Restaurant (or CafeOrCoffeeShop, BarOrPub): name, address, phone, hours via openingHoursSpecification, servesCuisine, priceRange, acceptsReservations and a menu link via hasMenu.
Menu with MenuSection sections and MenuItem entries: name, description, price in an Offer, dietary labels via suitableForDiet.
FAQPage for questions about kids, pets, deposits and reservations.
Event for tastings, concerts and other dated events.

Types and properties are documented on schema.org. Keep the markup in sync with the text: if you change a price on the menu, change it in the code too.

While you're at it, make sure robots.txt doesn't block AI crawlers (GPTBot, OAI-SearchBot, PerplexityBot, ClaudeBot) and that your menu and contact pages are in sitemap.xml.

GEO checklist for restaurants and cafes

Menu as an HTML page with prices and dietary labels.
Location page: address, landmarks, hours, kitchen hours.
One-click booking plus a phone number.
A private events page, if you host them.
Average check shown as a range.
Complete Google Maps profile with fresh photos.
Guest reviews, with replies.
Identical name, address, phone and hours on every platform.
JSON-LD: Restaurant, Menu, FAQPage.
AI crawlers allowed in robots.txt, sitemap.xml up to date.

How GEO differs from regular SEO is covered in a separate guide. To see how your restaurant's site looks to AI right now, run the free AI visibility audit: 8 technical factors in 60 seconds in a Telegram bot. For ongoing work, from setup to 4 new pages a month and a monthly before-and-after report, there's GEO Autopilot.

Mentions in AI answers usually appear within 2–4 months. Nobody can guarantee a specific spot in ChatGPT, which is why we measure results every month instead of promising them upfront.

FAQ

Is an Instagram page enough without a website?

Instagram helps, but it's an awkward source of facts: the menu, hours and prices are scattered across posts and stories, and some content isn't visible without logging in. A website with a text menu and a location page gives all that data one clear address. If you don't have a site yet, at least start with a fully completed Google Maps profile.

Our menu changes often. How do we keep it current?

Split the menu into a permanent part and seasonal sections so you only update those. Show a last-updated date and change prices everywhere at once: on your site, in Google Maps and in delivery apps. An outdated price in an AI answer lets down both the guest and you.

Do we need Menu markup if the menu is already on the page?

The text on the page is already useful. Menu and MenuItem markup also tells machines exactly where the dish, the price and the dietary label are. It doesn't guarantee a spot in answers, but it removes guesswork when the page is read.

What should a cafe chain with several locations do?

Create a separate page for each location with its own address, hours, phone and map. Each location should also have its own Google Maps profile. That way an AI assistant can recommend the cafe closest to the guest, not just the brand in general.

When will ChatGPT start mentioning our restaurant?

First mentions usually appear 2–4 months after the site, maps and reviews are in order. A lot depends on local competition and on whether guides, blogs and roundups write about you. Nobody can honestly promise a specific spot in an answer.