How restaurant groups choose the next neighbourhood to open in

Restaurant groups that open well rarely pick a neighbourhood on instinct alone. They shortlist areas where many people live or pass by for each restaurant already there, check what kind of restaurants those are and how well they are rated, and then test how easy the street is to reach at the hours they plan to trade. A neighbourhood that scores well on all of these is worth a visit; one that scores well on only one is usually a trap.

This article is for owners, expansion leads and operations directors of restaurant and cafe groups with a handful of sites or a few hundred. It walks through the measures that matter, how to read them, a worked example, the mistakes that cost groups a lease, and how to have an AI employee rebuild the shortlist every month.

The four questions behind every shortlist

• How many people does each restaurant already serve? Residents divided by the restaurants around them. A high number means demand that nobody is meeting yet; a very low one means a crowded street where a newcomer has to take guests from someone else.
• Who are the neighbours? A street of independents with strong ratings is a dining destination. A street of chains is a convenience strip. Both can work, but they call for different concepts and different rents.
• Is the area busy at your hours? A lunch concept needs people present at midday; a late kitchen needs evenings and weekends.
• Can guests and delivery riders get there? The area a kitchen can reach in ten or fifteen minutes decides both its walk-in catchment and its delivery zone.

What the figures look like in three cities

Big-city averages hide a lot, but they show how different markets are before you zoom in. xMap's market pages count the restaurants within 10 km of each city's centre from xMap's places database, and work out the mix of chains, ratings and openings from the few hundred nearest the centre.

Restaurants within 10 km of the centre of each city, xMap's places database, November 2026. Chains, independents, ratings and openings are counted among the restaurants nearest the centre.

Four things stand out. London is by far the biggest market, with 3,792 restaurants listed against 2,125 in Toronto and 1,640 in Sydney. Of the restaurants nearest Sydney's centre, 97% are independent, so a new chain concept there stands apart, and so does every weakness it has. London's nearest restaurants are rated higher on average (4.4), which raises the bar a newcomer is judged against. Toronto's nearest saw the most openings in the twelve months before November 2026 (22 of 290, against 19 of 324 in London and 9 of 303 in Sydney), a sign of a market still moving. None of this tells you which street to choose, but it tells you what kind of fight you are walking into. See the full pages for London, Sydney and Toronto.

Reading residents per restaurant

Population is counted on a grid of small hexagons, each about a tenth of a square kilometre, so a city can be read block by block rather than district by district. For each hexagon you add up the residents within a kilometre and count the restaurants in the same circle; dividing one by the other gives residents per restaurant.

Treat the number as a ranking, not a forecast. A block with three times the residents per restaurant of its neighbours deserves a look; it does not promise three times the covers. Offices, universities and stations bring people who do not live there, so a central business district can look empty on residents and still be full at lunch.

A worked example in Lyon

The figures in this example are made up to show the method. Say a group with eight bistros in Paris wants its first site in Lyon and has three candidate neighbourhoods.

A made-up example, not real figures for Lyon.

Neighbourhood C has the most residents per restaurant, but most of its few restaurants are chains along a ring road: people there eat out near work, not near home. Neighbourhood A is crowded and well served. Neighbourhood B has room, a strong independent scene to sit alongside, and enough residents to fill a bistro on weeknights. B goes on the visit list first, and the team checks how busy its streets are at dinner time before signing anything.

Mistakes that cost groups a lease

• Using one big circle for the whole city. A 10 km figure describes a market, not a street. Shortlist at the block level.
• Counting every restaurant as a rival. A ramen bar and a steakhouse rarely share guests. Pick the types of place that compete with your concept.
• Reading high ratings as saturation. Well-rated neighbours often mean diners travel to that street, which helps a good newcomer.
• Ignoring the hours. Footfall is counted as people-hours of presence, not visitors, and it arrives about ten days late. Use it to compare when areas are busy, not to count guests.
• Deciding once. Openings and closures change a street within months. Rebuild the shortlist regularly.

Letting an AI employee rebuild the shortlist every month

In xMap Flows the same method runs on its own as an AI employee, and the shortlist lands in your inbox with a map. A typical setup, in order:

1. A Schedule runs it once a month.
2. An Area of interest names the city.
3. Demographics returns every hexagon with its residents and age groups.
4. Nearby places, set to each incoming place with counts, counts your rival types within 1 km of each hexagon.
5. A Calculated field works out residents per restaurant, and Sort and limit keeps the 20 most underserved blocks.
6. xMap LLM scores each block from 0 to 100 against your concept, with a one-line reason.
7. Show on a map and an Email send the shortlist to the expansion team.

For the finalists, add a Travel area to see what each site reaches in ten minutes by car or on foot, and Mobility Footfall to see the busiest hour and day around it where xMap's mobility data covers the city. The ready-made delivery kitchen and next cafe templates follow this pattern and can be changed to your concept in one sentence. When two or three streets are left, Compare sites puts them side by side.