The Wonder Agency
For twenty years, being found online meant ranking on Google. Someone typed “Italian restaurant Marylebone”, got a page of results and a map, and chose. That still happens. But more and more people now skip the list and ask an AI assistant instead: “Where should I take my parents for dinner in Marylebone on Saturday?”
The answer they get back is different in one important way. It isn’t ten links. It’s a short reply with two or three names in it, and most people go with one of them. If your restaurant isn’t named, that guest never sees you.
With AI search there is no page two. You’re in the answer, or you’re not.
How AI assistants decide who to recommend
Assistants like ChatGPT, Gemini and Perplexity build their answers from what they can read about you across the web. No one outside those companies knows exactly how each one weighs things, but the patterns are clear enough to act on. They lean on:
- Your own website: what you serve, where you are, your opening hours, your prices and what kind of occasion you suit.
- Your Google Business Profile and other listings, and whether the details match everywhere.
- Reviews: how many, how recent and what people say about you in their own words.
- What others write about you: guides, press, blogs and local round-ups.
- How clearly all of that answers the question being asked, such as “romantic”, “good for groups”, “open late” or “dog friendly”.
In other words, AI recommends the restaurants it understands best and has the most consistent, trustworthy information about. That is good news for independents, because it rewards being clear and well described rather than having the biggest budget.
Why many restaurants are invisible to AI
In our audits, the same problems come up again and again. Menus are uploaded as PDFs or images that are hard for software to read. Websites say a lot about atmosphere and very little about facts: cuisine, neighbourhood, price, opening hours, whether you take bookings. Listings disagree about the hours. Reviews go unanswered. And nothing anywhere says what the restaurant is actually good for.
None of this is a problem for a regular who already knows you. It is a problem for an assistant trying to decide whether you’re the right answer for “a special occasion near Regent’s Park”.
What to do about it
- Put your menu on your website as a proper web page, not only a PDF.
- Say plainly what you are: cuisine, neighbourhood, price range, hours, and the occasions you’re best for.
- Make your details identical on your website, Google, booking platforms and listings.
- Ask happy guests for reviews, and reply to them, good and bad.
- Get written about: local guides, press and creators all add to what AI can find.
- Add structured data to your website so search engines and AI can read your details reliably.
Most of this is ordinary good practice. The difference is that it now decides whether you’re recommended at all, not just where you rank.
Measure it, don’t guess
You can’t improve what you can’t see, and you can’t see AI visibility by searching for yourself once. We built Wonder Score to measure it properly: it asks AI assistants the questions your guests ask, records how often you’re named against your competitors, and checks how readable your website is to AI. Then it tells you what to change.
If you’d like to know where you stand, run a free scan, or ask us for a free audit and we’ll look at your search and AI visibility alongside the rest of your marketing.
