Courtyard · AI Visibility Index
The State of AI VisibilityRestaurants · 2026
AI will happily name a restaurant. Understanding one is another matter. We measured how well ChatGPT, Gemini, Perplexity, and Google AI Mode understand restaurants across the questions customers ask, from services to availability to price.
Being named isn’t being understood.
How well AI understands each, scored 0 to 1. It has the surface; the specifics fall away.
The depth of understanding
The deeper the question, the thinner AI's understanding
AI can tell a customer what a restaurant does, understanding of services scores 0.74 out of 1. But the questions that actually decide who gets chosen, the price range, what's on the menu right now, and whether they take a walk-in, are exactly where it falls away.
Across restaurants, price is the floor: understanding of what it costs scores just 0.06. AI has the surface and almost none of the specifics a customer weighs before booking.
How it answers
Even when AI answers, it rarely commits
Thin understanding shows up as hedging. Across restaurants, 63% of answers wouldn’t commit, “prices aren’t listed,” “you’d want to check with them,” “it may vary,” and only 37% gave a customer something direct to act on. A hedged answer just sends them back to find out for themselves.
- “prices aren't listed”
- “you'd want to check with them directly”
- “it may vary”
- “I couldn't find a specific figure”
- “appears to offer”
Where AI learns about you
Where the answer comes from
Where does AI get its answers about restaurants? To decide who to recommend, it reads the reputation layer, critics and reservation platforms like OpenTable, The Infatuation, and Michelin. Ask about a specific restaurant, and the mix flips: those third-party sources recede and the business’s own website takes over, because that’s the only place the specifics live.
The source AI leans on for the specifics is the one you own, and it’s usually the part that’s thinnest and least current: prices saved as images, availability inside a booking widget, a page nobody has touched in a year. How AI finds local businesses: the full source breakdown.
What it means
AI can name a restaurant. It can’t yet answer for one.
Across restaurants the finding repeats: AI has the surface, what you offer, but the specifics a customer weighs before booking, the price, the availability, what sets you apart, thin out to vague, uncertain answers. The restaurants that get chosen will be the ones AI understands well enough to speak to.
That understanding is something you can give it. Courtyard publishes and maintains a live, machine-readable knowledge base of a business, its services, hours, pricing, and availability, in a form ChatGPT, Gemini, Perplexity, and Claude can read and keep current. No tool can promise a given answer, but a complete, current source is the difference between AI answering for your business and sending the customer off to find out for themselves.
Methodology
How this was measured
The businesses
The 120 restaurants AI recommends most across 50 US and Canadian markets, scored with the AI Visibility Index's production depth pipeline.
The questions
For each business we run a buyer's journey, offerings, availability, pricing, fit, and differentiation, across ChatGPT, Gemini, Perplexity, and Google AI Mode, and score how well AI can carry someone toward booking.
Posture & sources
Every answer is classified as direct, hedged, or deflected, and every source AI cites is bucketed by type, so the score, the posture, and where the answer came from all line up.
This edition
A point-in-time snapshot from the week of July 10, 2026. The live, continuously updated view is the AI Visibility Index.
These figures are an observed measurement of what third-party AI assistants say, gathered by Courtyard. They are not a rating, certification, or endorsement of any business. Full methodology on the Index.