Crawl and venue identity
One canonical page per venue, accessible HTML, accurate hours, menu, contact details and Restaurant schema that matches the visible page.
Restaurant AI Search
Schmitdy runs AI search as a measured restaurant growth channel. We find the questions where your venues are missing, repair the first-party record, earn the outside proof engines rely on, and track the path from recommendation to booking or enquiry.
What the market says
These figures describe specific US surveys and one Schmitdy London measurement window. They do not prove demand for every venue or guarantee a result. Sources: DoorDash, BrightLocal, and Schmitdy's UK index.
The operating model
A restaurant is easier to recommend when an engine can retrieve a clean venue record, understand why it fits the question, confirm the claim elsewhere, and see current local consensus. We work the weakest layer first.
One canonical page per venue, accessible HTML, accurate hours, menu, contact details and Restaurant schema that matches the visible page.
Direct answers for real choices such as neighbourhood, price, dietary needs, access, group size, occasion and booking constraints.
Relevant editorial, guide, listing and community sources selected from what the engines actually retrieve for your city and cuisine.
A fixed prompt set, source use, answer position, AI-referred sessions, qualified enquiries and booked calls read together over time.
What we ship
We fix venue pages, menus, schema, canonical signals, internal links and listing conflicts. Multi-site groups get a location-level entity map, not one blended brand page.
We commission the right page, guide, data study or free tool only after mapping the prompt and source gap. Every material article receives deep source research, original visuals and independent review.
We target relevant publications, guides and creators that engines already trust. The pitch is built around a real editorial idea, never a request to add a logo to a list.
Questions we measure
Where should I book a quiet anniversary dinner near Soho for about £70 per person?
Which London restaurant can seat 12, has serious vegetarian options and step-free access?
Hawksmoor or Blacklock for a client dinner where conversation matters?
We also keep brand facts, local discovery and venue comparisons separate. One favourable screenshot is not a trend.
Operator, not category tourist
Marco helped take Disperse from roughly three people to more than 200 and raise $35 million while building construction AI. At SlideSpeak, he led product and go-to-market work for an AI presentation platform used by more than one million people each month, with organic search and GEO as core growth channels.
Commercial model
Before work starts, we write down the restaurant prompt set, baseline, measurement method, calendar window and two realistic results. The same rules apply for the full 30-day month.
Not a fit: restaurants without control of a canonical website and listings; teams expecting guaranteed recommendations or an immediate PR blast; or businesses whose main need is paid media, revenue management or a new booking stack.
per 30-day month
Measurement
How often the right venue is named across the fixed prompt set, by engine and topic.
Whether hours, menu, location, price and access facts are correct and traceable.
Which pages are retrieved and cited before and after each intervention.
AI-referred visits, diagnostic completions, qualified enquiries and booked calls.
FAQ
Start with evidence
Bring one venue, one city and the customer you most want to attract. Marco will show you the prompt, source and conversion gaps worth acting on first.