Hospitality (hotels and restaurants)
People looking for somewhere to stay often ask an AI assistant to suggest a few hotels. TrustInsider follows the guest all the way to the booking and shows you which amenities, which ratings and which booking sites are used to describe your property.
The first request
The person describes what they need and the assistant answers with three or four names. A property makes the shortlist for the amenities its sources attribute to it, and drops off it if those sources describe them badly.
Hotel requests are almost always about an amenity or an occasion: the pool, parking, pets allowed, being close to a trade fair or a concert. The assistant matches the request against what it reads on booking sites and in reviews. A suspended amenity that a booking site still lists, such as a closed spa or a shuttle that no longer runs, brings in guests with an expectation the property cannot meet.
A listing of a property declared as a sample, with the amenities the booking site shows: the spa appears open while it is closed for works, and the shuttle from the station has been suspended.
What to avoid: the assistant describes amenities the property no longer offers, because it reads them from an outdated listing. What to do: align the booking-site profiles and the Google listing, and declare the amenities with structured data on the website.
Two properties, one name
When two hotels have similar names, the assistant may summarise the reviews of one while talking about the other. The guest reads complaints that have nothing to do with you and books elsewhere without knowing why.
It happens when businesses with the same name are not clearly distinguished in the sources. The defence is to make the entity unique everywhere: full name, address, phone number and identifiers identical on the website, the listing and the booking sites, so that every rating stays attached to the right property.
A sample answer to the question about a hotel’s reviews: the first part concerns the right property, the second repeats a rating written about a namesake in another city.
QuestionWhat is the Hotel Belvedere like according to people who have stayed there?
AI assistant · sample answer
Guests appreciate the breakfast and the view over the harbour. Some, however, mention noisy rooms facing the main road, as reported by the reviews collected by a booking site.
The rating really does concern the property searched for.
The complaint was written about a hotel with the same name in another city, on a main road.
The booking site lists the two properties with almost identical names and no full address.
Where the assistant reads
The hotel’s website is rarely the main source: answers rely on booking sites, review platforms, guides and the travel press. What matters is what the pages the assistant actually reads say about you.
Each type of source has a different owner and a different pace. You manage the website and the listing; you update booking sites from their dashboards; reviews, guides and articles are written by others. The job is not to move the assistant onto your website, but to make the facts consistent where it finds them, starting with the sources that change most often and that you control least.
Five types of source that assistants read to describe a hotel, placed according to how directly the property manages them and how often they change, without numbers.
How often it changes
How much you manage it
It changes when you decide: it is the factual base, not always the source that gets cited.
You manage it, but it receives questions, photos and suggested edits from others.
Amenities, prices and availability are updated from the booking site’s dashboard.
Every stay can add a rating; you can only respond.
They rarely change and stay widely read for a long time, even when they are out of date.
Cycle after cycle
Correcting a listing is not enough: you need to see whether the answers change. The platform repeats the same questions every cycle and shows how the property’s AI Trust Score moves, next to that of competing properties.
The glimpse shows the AI Trust Score trend for a demo brand: one point for each complete survey of the same panel, with the 95% interval band and a competitor’s line calculated with the same method. Inside the interval a difference is not certain, and the page says so. For a hotel it is the way to tell whether removing the spa from the portal or correcting the shuttle on the listing really moved the answers, or whether the change is noise.
After the stay
After the stay comes the review, and the review goes into the answer the next guest will read. Responding on every platform the assistant reads counts as much as the score.
European rules on reviews require whoever publishes them to say whether and how they check that they come from real customers, and ban fake or bought ones. In Italy, hotels and restaurants may not offer discounts or benefits in exchange for a review. For a hotel this means two practical things: no shortcuts on ratings, and precise responses on the facts, with no promises, because the property’s response is also read and summarised.
Four sample steps: the guest writes a review, the property responds on the facts, the assistant summarises both, and the next measurement says whether the summary has changed.
A review reports a cold room in the morning.
It explains that the heating has been replaced, and when.
The summary includes both the problem and the fix.
At the next cycle you check that the sentence about the cold has gone.
Many properties, one brand
A chain lives the same conversation multiplied by every hotel. Amenities, opening times and star rating change from one property to another, and the assistant tends to spread across all of them what it reads about just one.
For a hotel group the typical risk is one hotel’s amenity attributed to another under the same banner, or one property’s seasonal closure extended to the whole brand. The location-by-location checks, the order in which to run them and the city view are the same as for any multi-location network, and they live on the dedicated page.
For three typical facts in a hotel chain: how the answer generalises them and where they should be written so they stay tied to the individual property.
| Fact | How the answer generalises it | Where it is kept separate |
|---|---|---|
| The pool | All the group’s properties have a pool | The page and the listing of each individual hotel |
| Seasonal opening | The chain closes in winter | The calendar of the property that actually closes |
| The star rating | All five-star | Each hotel’s classification, with the body that awards it |
The restaurant trade
A restaurant has its own questions and sources: opening hours, menus, allergens, food guides. That is why it has its own page, while this one stays on the hotel guest’s journey.
A restaurant is often chosen in a few minutes, with a spur-of-the-moment question: open tonight, suitable for someone with coeliac disease, near the hotel. Listings, table-booking sites and reviews give opening hours and menus in different ways, and the assistant picks which one to believe.
A sample request, “where can I have dinner tonight near the hotel”, and how the two businesses read it: the hotel as a service to its guests, the restaurant as a booking to win.
| For the hotel | For the restaurant | |
|---|---|---|
| What counts | That the in-house restaurant appears among the suggestions for people staying there. | That the venue shows as open tonight and within walking distance. |
| The decisive source | The amenities page and the property listing. | The venue’s listing and the table-booking site. |
| The typical mistake | The in-house restaurant described as open to non-residents too. | A closing day shown as open. |
Special properties
Some accommodation types have a problem all of their own, and for each one there is a use case designed for the person who runs it. We start with the farm stay, where the season changes the questions.
A farm stay is asked about for experience and season, a campsite for pitches and facilities, a wedding venue for capacity and free dates. The guest’s journey is the same as on this page, but the sources and the mistakes change: the use cases tell them from the owner’s point of view.
For three types of accommodation: the typical question the guest asks the assistant and the fact that decides whether the property makes it into the answer.
| Property | Typical question | The deciding fact |
|---|---|---|
| Agritourism | Where can I eat the farm’s own produce in the hills? | The season’s opening dates and what is actually produced |
| Campsite | Campsite with motorhome pitches and heated bathrooms? | The facilities of each area and the opening period |
| Wedding venue | A villa for a wedding with a hundred and twenty guests? | The declared capacity and the dates still available |
Frequently asked questions
Which sources the assistant uses to describe a hotel, what to do about a misreported amenity, how properties with the same name are kept apart and whether the measurement works for a group.
Mostly from booking sites, review platforms and the Google listing, more than from the property’s website. That is why an amenity corrected only on the website can stay wrong in the answer.
You correct the source that reports it: the booking-site profile, the listing, the amenities page with structured data. Then you check at the next cycle, on the same questions.
By making the entity unique: full name, address, phone number and identifiers identical on the website, the listing and the booking sites, and by reporting to the booking sites the listings that confuse the two properties.
Yes: it tracks the tone and themes of reviews and places them next to the assistants’ answers, so you see which ratings make it into the summaries.
Yes: every property has its own questions and facts, and the location view is the one used for multi-location networks.
Related paths
Each role uses the platform with a different goal. From here you compare the other solutions, the path from brand to verification and the methodology behind the numbers.
Check how ChatGPT and Gemini describe your restaurant. Compare opening hours, menus and services with your website and booking platforms.
ReadLuxury hotels: services, awards and distinctive features of the property in AI answers, with the checks to make before quoting an accolade.
ReadHoliday rentals and B&Bs: house rules, check-in, amenities and location of the property in AI answers, to check before asking for a correction.
Read