How to get your restaurant recommended by ChatGPT, Gemini and AI Overviews

There's no sign-up or special trick for getting recommended by ChatGPT, Gemini or Google's AI Overviews. They build answers from search indexes, map listings and partner data, so the work is local search groundwork: the same name, address and phone everywhere, complete Google, Bing and Apple listings, a steady flow of genuine reviews, and a website with your menu, hours, prices and booking link as text.

#Where do ChatGPT, Gemini and other AI assistants get restaurant information?

Each company documents part of this and keeps the rest to itself. What follows is only what their own pages say, checked October 2026.

#ChatGPT

OpenAI's help article on searching the web with ChatGPT says it uses your location to find local results, including restaurants. "ChatGPT search sometimes partners with other search providers", and the provider privacy policies it links to are Microsoft's and Shopify's. It may also share your location with "trusted third-party providers that help provide more local information, such as nearby business listings", which it doesn't name.

Bookings are more specific. ChatGPT can show availability from OpenTable, Resy or Yelp, and the same article says only OpenTable works globally (Resy is US only, Yelp US and Canada). For a UK venue, OpenTable is the one that can put bookable times inside a ChatGPT answer.

For your website, OpenAI says to allow its search crawler, OAI-SearchBot, and to make sure your host or content delivery network lets its published IP addresses through. And, in OpenAI's words, "Placement is not guaranteed."

#Google AI Overviews, AI Mode and Gemini

Google's page on AI features and your website is unusually blunt: "There are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary." A page has to be indexed and eligible to show in Search with a snippet. Its best practices include keeping important content in text, making structured data match the visible text and keeping your Business Profile up to date.

For Gemini, Google's help on connected apps lists Google Maps among the services whose public information Gemini can use. Inside Maps itself, Ask Maps launched in March 2026 in the US and India, drawing on information about more than 300 million places and reviews from over 500 million contributors. My reading: for Google's assistants, your Google Business Profile carries a lot of the weight.

#Microsoft Copilot

Microsoft's support page on how web search works in Microsoft 365 Copilot Chat (the work version) says Copilot writes a short query from your prompt and sends it to Bing. In Bing itself, Microsoft's privacy statement describes Copilot Search as reviewing and summarising search results from across the web. Business listings in Bing come from Bing Places for Business, which is free, shows listings in Bing search and Bing Maps, and lets you import from your Google profile.

#Perplexity

Perplexity's help centre says it searches the internet in real time and cites its sources. Its crawler documentation says PerplexityBot is what surfaces and links websites in its results, and recommends allowing it in robots.txt. I couldn't find any Perplexity documentation saying where its local business listings come from.

#Siri and Apple Maps

Apple replaced Business Connect with Apple Business on 14 April 2026, free, with existing claimed locations moved across automatically. You claim your location and manage a place card (photos, location details, hours) that Apple says displays across Apple Maps, Safari and Spotlight. Its announcement also lists Mail, Wallet and Siri among the places the platform helps businesses reach customers. Separately, Apple says its crawler, Applebot, powers the search behind Spotlight, Siri and Safari.

#What nobody outside these companies knows

None of them publishes how it decides which venue to recommend for "best pub for Sunday lunch near me". OpenAI says ChatGPT ranks results "using multiple factors" and that placement isn't guaranteed. Google says no special optimisation is needed. Microsoft and Perplexity say very little about local ranking at all.

Answers also change from person to person. OpenAI explains that ChatGPT rewrites questions using your approximate location and, if memory is on, things you've told it before, so its example vegan user in San Francisco gets "good vegan restaurants San Francisco". Asking ChatGPT once whether it recommends you tells you very little.

So I'd be wary of anyone selling guaranteed AI rankings. What you can control is being the clearest, most consistent source of facts about your venue in every place these systems read.

#Make your name, address, phone and listings match everywhere

Write down exactly how your venue's name, address and phone number appear, then make every listing match it: Google, Bing, Apple, Facebook, Tripadvisor, your booking platform, delivery apps and your own website. "The Brass Heron" on Google and "Brass Heron Pub & Kitchen" on Tripadvisor, with two different phone numbers, is the kind of mismatch to hunt down. If Google says you close at 11 and your website says midnight, you're asking a machine to guess.

Then work through the three big listings:

1. Fill in every field of your Google Business Profile, set special hours and keep it current, since Google's AI features page names it directly. Our Google Business Profile guide covers it step by step.
2. Set up Bing Places by importing from Google; Bing's announcement says the new import carries over the name, hours and contact details more reliably.
3. Claim your location in Apple Business so your place card shows the right hours and photos in Maps, Safari and Spotlight.

If you take bookings through OpenTable, treat that listing as a fourth, given it's the provider ChatGPT can book in the UK.

#Why do reviews matter for AI recommendations?

Reviews are one of the few signals any of these companies names. Google's local ranking page says prominence is partly based on how many reviews you have, and that "more reviews and positive ratings can help your business's local ranking." Ask Maps, by Google's own description, draws on reviews from its contributors.

When an assistant calls somewhere "good for groups" or "a bit slow on Fridays", that description has to come from somewhere, and reviews are the obvious candidate. Nobody outside these companies can confirm that, but it's one more reason to keep honest reviews coming in and to reply to them. Our guide on how to respond to restaurant reviews covers the replies and the rules (UK law has banned fake reviews since April 2025, and the CMA's fake reviews guidance explains what that covers).

#Put your menu, hours, prices and booking link on your website as text

Google's guidance for AI features is to make sure "important content is available in textual form", and a photo of your printed menu isn't text. Nor is a homepage graphic saying "Kitchen open till 9". If the only place your Sunday menu exists is a JPEG on Instagram, you're relying on every system to read an image correctly (sometimes very confidently, and wrongly).

What I'd put on the site as plain text:

• The full menu, with prices and dietary labels written out (vegan, gluten free), on its own page.
• Opening hours and kitchen hours, including how you handle bank holidays.
• Address with postcode, phone number and a map link.
• A booking link that goes straight to your booking system.
• The facts people ask about: dogs, children, step-free access, garden, screens for sport, private hire, parking.
• An events page with dates written as text.

Each venue in a group should get its own page with its own address and hours. When the menu changes, update the page the same day; our guide to launching a new menu has a running order for that.

#Add Restaurant or BarOrPub schema markup

Schema markup is a few lines of code that label your venue's facts for machines. Schema.org has a Restaurant type and a BarOrPub type, both under FoodEstablishment, with properties such as servesCuisine, hasMenu and acceptsReservations.

Google's local business structured data documentation lists name and address as required, recommends telephone, opening hours, servesCuisine, menu, priceRange, url and geo, and tells you to use the most specific type you can. It also says Google doesn't guarantee that features using structured data will show up. Its general guidelines say not to mark up anything readers can't see, so every fact in the markup should also be on the page.

A trimmed example for a fictional pub:

{
  "@context": "https://schema.org",
  "@type": "BarOrPub",
  "name": "The Brass Heron",
  "url": "https://www.example.com/",
  "telephone": "+44 1632 960123",
  "address": {
    "@type": "PostalAddress",
    "streetAddress": "12 Mill Lane",
    "addressLocality": "Anytown",
    "postalCode": "XX1 1XX",
    "addressCountry": "GB"
  },
  "openingHoursSpecification": [
    {
      "@type": "OpeningHoursSpecification",
      "dayOfWeek": ["Monday", "Tuesday", "Wednesday", "Thursday"],
      "opens": "12:00",
      "closes": "23:00"
    },
    {
      "@type": "OpeningHoursSpecification",
      "dayOfWeek": ["Friday", "Saturday"],
      "opens": "12:00",
      "closes": "23:59"
    },
    {
      "@type": "OpeningHoursSpecification",
      "dayOfWeek": "Sunday",
      "opens": "12:00",
      "closes": "22:30"
    }
  ],
  "servesCuisine": "British",
  "priceRange": "££",
  "menu": "https://www.example.com/menu",
  "acceptsReservations": "https://www.example.com/book"
}

Google's documentation doesn't list acceptsReservations, but it's part of schema.org's FoodEstablishment type. Most website builders and plugins can add markup like this; if someone else runs your site, send them this section.

#Check you're not blocking the crawlers

Make sure your robots.txt file lets search crawlers in. Google describes robots.txt as a way to manage crawler traffic, and says it's not a mechanism for keeping a page out of Google. It also says robots.txt rules for Googlebot are the control site owners have over AI Overviews and AI Mode.

The crawlers that decide whether you can be found:

• Googlebot, for Google Search, AI Overviews and AI Mode.
• Bing's crawler, for Bing and Copilot.
• Applebot, for Spotlight, Siri and Safari.
• OAI-SearchBot, for ChatGPT search (OpenAI's crawler page).
• PerplexityBot, for Perplexity.

Blocking AI training is a separate decision with separate names. OpenAI says GPTBot crawls content that may be used to train its models. Google says its Google-Extended token "does not impact a site's inclusion in Google Search nor is it used as a ranking signal." Apple says disallowing Applebot-Extended opts you out of training while your content stays discoverable in Spotlight, Siri and Safari.

The simplest robots.txt that lets every crawler in is two lines:

User-agent: *
Allow: /

To opt out of training only, add these groups underneath:

User-agent: GPTBot
Disallow: /

User-agent: Google-Extended
Disallow: /

User-agent: Applebot-Extended
Disallow: /

Bot protection on your hosting or security service can still block crawlers that robots.txt allows, which is why OpenAI and Perplexity both publish their IP addresses; ask whoever runs your site to check. Both also say robots.txt changes take about 24 hours to be picked up. The wider local picture is in our guide to local SEO for pubs and restaurants, and using AI to write your posts is a separate question, covered in can AI write your restaurant's social media?