How to list restaurants by cuisine in a city, and what the labels miss
Cuisine place types make it possible to count Italian, Thai or Korean restaurants across a city. What each type covers, where the labels fall short, and how to fix it.
Part of the guide Location intelligence automation
To list restaurants by cuisine in a city, search a places database for the cuisine types one by one, district by district, and then deal with the restaurants that carry only the plain label Restaurant. The first part is mechanical. The second is where most cuisine counts go wrong, because a large share of places are never given a cuisine at all.
This guide is for food suppliers who sell one kind of ingredient, delivery platforms looking for gaps, landlords filling a food court, writers compiling guides and anyone opening a restaurant who wants to know what the street already has. It explains how cuisine types are organised, what they cover and miss, a made-up worked example and an AI employee that keeps the picture current.
How cuisine types are organised
xMap's places database gives every place one type out of more than 5,000. For food, the types are grouped by cuisine, and the groups are very uneven in depth.
| Group | What is inside |
|---|---|
| Italian | 13 types: Italian restaurant plus regional ones such as Neapolitan, Sicilian and Tuscan |
| Japanese | 40 types, from Izakaya and Tempura to Yakitori and Tonkatsu. Sushi and ramen are separate groups |
| Chinese | 19 types, including Cantonese, Sichuan and Shanghainese. Dim sum, dumplings and hot pot each stand alone |
| Indian | 32 types, with South Indian as its own small group |
| Korean | 26 types, many named after one dish |
| Mexican | 7 types, with Tex-Mex apart |
| Thai, Peruvian, Ramen | One type each |
| Restaurant | 137 types: the plain label, plus many national cuisines without a group of their own, such as British, Belgian and Croatian |
A sample of the cuisine groups in xMap's places types.
Three things follow. A cuisine is rarely one type, so searching Japanese restaurant alone misses the izakayas. Some foods you might file under a cuisine are their own group: pizza places are not in the Italian group, and kebab shops are not in the Middle Eastern one. And the catch-all Restaurant group is where everything unlabelled ends up.
What the labels cover and miss
- One type per place. A restaurant that serves both sushi and Korean barbecue has one label, not two. Fusion kitchens land wherever the labeller put them.
- The plain Restaurant label. Many places, especially independents, are typed simply Restaurant. They may be the best Italian in the city. A count by cuisine types alone leaves them out, so it is a floor, not a total.
- Dish types versus cuisine types. A place typed Biryani restaurant is in the Indian group; a place typed Noodle shop could be almost anything. Read the group, not just the word.
- Diet labels. Vegetarian, vegan, halal and gluten-free are types too, and a place carries one of them instead of a cuisine, not as well as.
So be careful how you report the result. It is fair to say that the database lists a number of places typed as Thai restaurant in an area. It is not fair to say that this is how many restaurants serve Thai food there.
Searching a city without holes
In Nearby places, one search is one type around one point: the 500 nearest, up to 25 km away. A block takes up to 12 types. A small group, such as the seven Mexican types, can be picked whole. A deep one, such as Japanese, has to be trimmed to the types that matter or spread over several blocks.
Split the city so that no search hits its limit. Put districts in an Area of interest, up to 8 a run, and each is searched from its middle to its edge with only the places inside kept. Every restaurant then carries its district, which is exactly the column a cuisine table needs. The article on listing all the restaurants in a city covers the splitting in more depth.
A made-up worked example
These figures are invented. A wholesaler of Korean ingredients wants to know where Korean restaurants are in six districts of Sao Paulo.
- Nearby places searches the Korean types it cares about (Korean restaurant, Korean barbecue restaurant and a few dish types) in the six districts. Say it finds 58.
- A second Nearby places block searches the plain type Restaurant in the same districts. Say 2,300 come back across the six.
- Filter keeps those whose name has a telling word, such as korean, seoul, bibimbap or kimchi. Say 21 pass.
- xMap LLM reads the name of each of the 21 and marks whether it is likely to be Korean, with a reason. Say 17 are kept.
- Remove duplicates joins both lists: 75 restaurants, of which nearly a quarter had no cuisine label.
The lesson is in the last number. Without the second pass, the wholesaler's sales team would have missed 17 customers. The name check is a guess, and the flow should say so: mark those rows as found by name so a person can confirm them on the first call.
Reading the result
Beyond the type, each restaurant comes with its rating and number of reviews, brand and whether it belongs to a chain, address, phone and website, and, when known, a dining type and a price level. That is enough to answer the questions people really have:
- Which districts have many restaurants and none of this cuisine?
- Is this cuisine mostly chains or independents here?
- Where are the well-rated ones clustered, and where is the rating low?
An AI employee that keeps the table current
Cuisine maps shift as places open and close. A flow can redo the count on a timetable, which is the everyday use of location intelligence automation:
- Schedule on the first of each month.
- Area of interest with the districts, then Nearby places with the cuisine types.
- Only new to pass on restaurants not found before.
- Make a report or an email with the new places by district and cuisine.
The template Cuisines missing in each district is built for the gap question, and Restaurants worth reviewing this month for the well-rated newcomers. Either can be changed to your city and cuisines in a sentence.
Questions
Can I get a list of all the Italian restaurants in a city?
You can list every place typed as an Italian restaurant, including regional types such as Neapolitan or Sicilian. Places typed only as Restaurant are not included unless you add a second pass by name, and pizza places are a separate group.
Why is my cuisine count lower than I expected?
Because many restaurants carry the plain label Restaurant with no cuisine. A count by cuisine types is a minimum. Checking names in the plain group recovers some of the rest.
Does a restaurant have more than one cuisine label?
No. Each place has one type, so a fusion restaurant appears under a single label chosen for it.
How many cuisine types can one search cover?
One Nearby places block takes up to 12 types, each searched separately in each area. Large groups such as Japanese or Indian need trimming or several blocks.
How often does the data change?
xMap's places database is refreshed every month. A monthly flow with Only new reports the restaurants added since the last run.
Templates that do this
Read next
More on business and places data · Every article · Templates