For restaurants
The AI answer layer for restaurants
Diners ask an AI where to eat, what is on the menu and whether you can feed them safely. LBOX measures what the engines say about your restaurants, finds where a review site or a delivery app is answering in your place, and builds the evidence that changes it.
Measured across
Perplexity
Gemini
Google AI
Claude
Copilot
Meta
DeepSeekGrok... and moreEvery one of them is competing for the same answer. One name comes back when a buyer asks.
What makes restaurants different
Not general marketing advice. These are the constraints your category actually operates under.
Your menu is a picture, so the model cannot read it.
Most restaurant menus live as a PDF or a flat image, sometimes inside a booking widget. An engine asked what you serve, what a dish costs or whether you do a set lunch has nothing to read, so it answers from a review someone wrote two years ago.
Delivery marketplaces answer the order question and charge you for it.
Ask an engine how to order from you and it surfaces DoorDash or Uber Eats before your own page. You pay a commission on that order, and the platform that took the cut is also the source that decided you were worth mentioning.
Allergen and dietary answers come from strangers.
Gluten free, nut free, halal, vegan. A model that tells a coeliac diner you can accommodate them is making a promise on your behalf, from a third-party listing, and your kitchen is the one that has to keep it.
Best near me is settled before anyone reaches your site.
The engines resolve where to eat through maps profiles, review counts and a handful of city guides. By the time a diner would have visited your website, the shortlist already exists and you are either on it or you are not.
What you get back
The console, measuring a restaurants company against the questions its buyers ask. Every brand named, every domain cited, every engine separately.
A worked example on an illustrative company, shown to display the console. Your own numbers come from measuring your questions through the engines’ own APIs.
See where the engines send diners
Visibility and position on the unbranded where-should-we-eat questions diners ask in your area, measured per engine.
- Ranked against the restaurants the engines actually named
- See where a model reads your pages without naming you
You against the field, engine by engine
unbranded questions| Brand | Gemini | AI Overviews | OpenAI | Perplexity | Claude | ChatGPT |
|---|---|---|---|---|---|---|
| The Copper Larder | 32% | 29% | 26% | 22% | 14% | 30% |
| Marlowe House | 26% | 23% | 21% | 18% | 11% | 24% |
| Cedar & Ashyou | 21% | 11% | 18% | 15% | 8% | 4% |
| Ottoline | 12% | 9% | 9% | 11% | 5% | 10% |
Turn what the answers get wrong into work that ships
Findings arrive as a worklist sorted by who can act: what LBOX writes, what needs your decision, what needs your kitchen or operations team.
- Scored against your cuisine and city, not a generic SEO checklist
- Nothing publishes without your sign off
The worklist
Prove the change, service after service
Every measurement is stored, so the line lengthens each cycle and the movement is on the record rather than asserted.
- The same questions, asked again and checked
- Sources broken out per model, per page
Visibility over time
every measured runKnow when the models come to read your menu
Crawler activity captured at the edge, so you can see which models fetch your menu, location and booking pages.
- Every model, every page, logged over time
- Robots and content signals kept current
Crawler activity
hits by model, last 30 daysWhere you disappear as the diner gets closer to booking
Presence falls away as intent rises. The book-a-table and order-from questions are the ones that convert, and the ones the platforms win.
- Visibility split by where the question sits in the journey
- Names the stage that costs you the cover
Visibility by buying stage
where you show up in the journeyThe sources the engines trust about restaurants
Maps profiles, review platforms, booking sites and delivery apps decide what the engines repeat about you. Your own menu is barely in the conversation.
- Owned versus earned, so you know what to build and what to earn
- Every citation traced to the answer it supported
Which domains the engines cite
668 citationsThe restaurants program
Everything below is included. Measured through the engines’ own APIs, so the same questions can be asked again and the result checked.
- Measure what every major AI engine says about your restaurants today, through the engines own APIs
- The diner questions you are actually asked, mapped by cuisine, occasion, dietary need and neighbourhood
- Where a review platform or delivery app is answering in your place
- Whether the engines describe your menu, hours, locations and booking options correctly
- Which sources the engines cite for your category, and which of those you control
- Menu, location and dietary pages built as text the engines can actually read
- Re-measurement on the same questions, so the change is provable rather than asserted
What this does not show: what any individual user saw. LBOX measures what the engines answer through their own APIs, which is repeatable and checkable. It is not a record of any one person’s session, and it does not predict revenue.