GEO for Hotels, Restaurants and Cafés
Guests ask an assistant where to go; your menu, your hours and your booking route have to be published so they can enter that answer.
Answering "is there a restaurant with a vegetarian menu in Beşiktaş" — a district of Istanbul — the assistant lists places whose menus it can read. In most Turkish businesses the menu is one image file: a person looks and reads, a machine looks and sees nothing. The vegetarian menu may be excellent; the list answering the question will not know.
The same repeats with opening hours, the booking route and dietary information. If only the sign on the door knows you open on Sunday morning, you cannot enter the "open for breakfast on Sunday morning" answer. Same on the hotel side: parking, early check-in and the children's policy are mentioned somewhere, never in one tidy machine-readable place.
The work is publishing every practical detail behind a hospitality decision — menu, hours, reservations, location, facilities — as readable text and structured data, with a routine that keeps it current. The most expensive mistake here is a guest at a closed door because the hours were wrong. The second is promising a dish not on the menu; both are data freshness problems, not creative ones.
What your customer asks the assistant
What we fix in this sector
Menu into text and schema
The image menu becomes real text: sections, dish names, short descriptions and prices on a readable page, marked up with Menu, MenuSection and MenuItem. Even if the PDF or JPG stays, the text version sits beside it — that is the quotable one.
Dietary, allergen and cuisine tags
Each dish gets its tags — vegetarian, vegan, gluten-free, halal — through suitableForDiet, with main allergens written plainly. Cuisine type is set with servesCuisine. No tag goes on unconfirmed; a gluten-free mark not checked with the kitchen is hard to undo.
Opening hours and the exception calendar
Weekly hours are published as openingHoursSpecification, with separate records for breakfast and dinner service. Public holidays, annual closures and special-day hours go in as exceptions. The data syncs to the map profile; two places showing different hours is the commonest fault.
The reservation and ordering route
The reservation page is declared with potentialAction and acceptsReservations marked. Phone, WhatsApp and any booking link sit in one place; party size limits, how long a table is held and cancellation terms are in plain text, so the answer is on the page.
Location, transport and facilities
Beside address, coordinates and map link we add transport: minutes from which stop, how long from the airport, parking and whether it is paid. On the hotel side parking, pool, pets and early check-in become amenityFeature, check-in and check-out times their own fields.
Seasonal content and the review flow
Demand tied to times of year becomes pages in advance: Ramadan iftar menus, New Year programmes, the summer terrace, holiday hours. Each seasonal page has planned open and close dates. Reviews get a response flow, and repeated complaints about menu, hours or reservations route into fixing the page.
RestaurantStructured data for this sector
Restaurant and café pages use the Restaurant type; under FoodEstablishment, it carries fields for menu, cuisine, reservation acceptance and price range. The menu attaches with hasMenu and splits into MenuSection, MenuItem and suitableForDiet, so a dietary question is verifiable from structured data as well as text.
Hotel pages use Hotel or LodgingBusiness. Room types, check-in and check-out times, facilities and pet policy use that type's own fields; a hotel's restaurant attaches as a separate Restaurant record.
The rule is the same everywhere: only information written on the page and confirmed by the business gets marked up. Unverifiable fields — ratings, awards, capacity — stay empty. A dish off the menu or a changed hour is updated in schema along with the page; the two drifting apart turns quickly into misinformation.
What we measure
- A fixed list of location, dietary and hours questions goes to the assistants regularly; whether the business appears is recorded.
- Traffic to the menu page and its click-through to the reservation step is tracked.
- On Google Business Profile, searches, direction requests, phone and website taps are followed.
- Reservation channels are split: your own site, phone and third-party platforms show separately.
- Review count, average rating and recurring themes — hours, waiting, menu — are classified monthly.
- Whether hours match across site, map profile and reservation platform is checked periodically.
Frequently asked questions
Enough for a person, not a machine. Dish names, prices and dietary details in a PDF or image menu are not read reliably by crawlers, and even then the section-to-price link is lost. A text version makes the menu searchable and a source for dietary and cuisine questions.
You don't, but "what would two people pay here" is asked constantly and a page with no price cannot enter that answer. Short of a full list, a range and a few examples help. Every published price needs a named owner; a stale price is worse than none.
We set up a single source you manage hours from: site, map profile and reservation platform take the same data. Known exceptions — holidays, annual closure, special days — go into the calendar in advance. On a sudden change the map profile updates first, because it spreads fastest.
The platform is good at taking bookings, but holds menu, transport, facility and policy information in its own format, and that page accrues to the platform, not you. Your site becomes the permanent source while bookings still route there. Both showing the same hours and menu is enough.
Reviews are a read and quoted source in hospitality; a recurring complaint matters far more than any single bad review. So we do not only reply, we classify. If waiting times or a menu misunderstanding recurs, the fix is correcting the page, not writing a reply.
Hotels, Restaurants and Cafés — let us look at where you stand
We check how AI assistants answer for your brand today, then work out which of the items on this page are missing on your side.
Get in touch