How your restaurant shows up when people ask AI where to eat.

The assistant answers from what it can read

Somebody types "good breakfast near the train station, somewhere I can bring kids." The assistant doesn't eat anywhere. It answers from what it can find and read about the restaurants nearby: their own websites, their Google listings, review sites, local articles and menus. If your facts are clear and they agree with each other everywhere, you're easy to recommend. If your site is a photo of a menu from 2019 and your hours disagree with Google, the assistant either skips you or gets you wrong.

That's the whole game, and it's good news for an independent restaurant. You don't need a big budget to be the clearest source of facts about your own place.

What the assistant needs from you

It needs the facts a diner would ask a friend about, written as plain text on a page it can read. Your name and address, your phone number, your hours including holidays, and what you serve, with prices. Then the questions that decide where a family or a group goes: parking, reservations, kids, the bar, outdoor seating, gluten-free and vegan options, wheelchair access, private rooms.

A menu that lives only in a PDF or a photo is hard for these systems to read reliably. So is a page that shows nothing until a script loads. Put the menu on the page as text, with prices, and keep it current.

Say it the same way everywhere

Your website and your Google Business Profile should agree on your name, address, phone and hours, and so should the big review sites. When they disagree, an assistant has to guess which one is right, and it might pick the old one. Fix the Google listing first, because so much reads from it, then make your own site match.

Mark it up so machines can read it

Search engines and AI systems read a format called structured data (you'll see it called schema.org markup). It's a block of code on each page that says, in a way a machine can't misread, "this is a restaurant, here's its address, these are its hours, this is its menu." Diners never see it. Most restaurant sites don't have it, or have it wrong.

Don't block the readers

Some website hosts and security services now block AI crawlers by default, and some old site settings tell every crawler to stay out. If an assistant can't read your site, it answers from everything else about you instead. Ask whoever runs your site whether AI crawlers, such as OpenAI's GPTBot and Anthropic's ClaudeBot, are allowed in.

Reviews in your guests' own words

Assistants pay attention to what diners say. Reviews that describe real dishes and real visits ("the eggplant parm is enormous, ask for the booth by the window") give an assistant something specific to repeat. Reviews collected on your own site, from guests who actually ate there, belong to you and stay on your page.

What AEO and GEO mean

You'll hear marketers say answer engine optimization (AEO) and generative engine optimization (GEO). Both mean the same work: making your restaurant the clearest, most trustworthy source of facts about itself, so AI systems repeat you correctly. There's no trick to it, and anyone selling a secret is selling you something else.

What Arara does about all of this

Every Arara site is built from facts you've confirmed, with your menu and hours as plain text. Each page carries structured data that tells machines where you are and when you're open. It also says what you serve and how to order. Arara asks you the questions diners ask, once, and puts the answers on your site. Your reviews come from your own guests through table-tent QR codes. And AI crawlers are welcome. See what's on an Arara site.