Site selection for a bakery: the morning trade, the walk-in and the wholesale run

Put a bakery where people pass on foot in the early morning, where enough homes sit within a short walk for the weekend trade, and where a van can load at dawn without a ticket. Then check what other bakeries are already on that route and whether they are chains selling on price or independents selling on quality. Most of a bakery's takings happen before ten, so the site must work at seven in the morning, not at the hour you visit it.

This article is for bakers opening a first shop, bakery chains adding a branch and cafes adding a bake-on-site counter. It looks at the three kinds of bakery trade, what the market looks like in three cities, a worked example in Melbourne, the mistakes, and an AI employee that checks candidate units.

Three kinds of trade, three kinds of site

• The commuter breakfast: coffee and a pastry on the way to work. It needs a pavement people walk at 07:00 to 09:00, close to a station, a bus stop or offices. Weekends are quiet.
• The neighbourhood loaf: bread, cakes for the weekend, orders for birthdays. It needs homes within a 10-minute walk and does its best days on Saturday.
• The wholesale bake: bread for cafes, hotels and restaurants. The shop window matters less than a kitchen big enough to bake at 03:00 and roads clear enough to deliver by 07:00.

Most bakeries mix all three, but one leads. Decide which before you look at units, because each points to a different street.

What the market looks like in three cities

xMap's places database lists 512 bakeries within 10 km of the centre of Toronto, 1,367 in London and 507 in Sydney (November 2026). Among the few hundred nearest each centre, the mix is very different:

xMap's places database, November 2026; the figures are of the bakeries nearest each city's centre.

In central London about a third of the nearest bakeries belong to a chain, against about one in ten in Toronto and Sydney. A new independent in central London is therefore more likely to share a street with a chain bakery selling on speed and price; it should compete on what the chain does not do. In Toronto, 28 openings among the nearest 366 in a year show a market still drawing new bakers in.

How to read a candidate street

1. Measure the morning. Mobility Footfall gives the area's footfall per day in people-hours of presence, the busiest hour and how the day splits into parts. A strong early-morning share is what a commuter bakery needs. The data arrives about 10 days late, so it describes recent weeks, not today.
2. Count the homes. Demographics gives residents within a walk, where the country has figures (among them Australia, Canada and France; not the United Kingdom or the United States).
3. List the bakeries on the route. Nearby places finds bakeries, cafes and supermarkets with an in-store bakery within a short walk, and says which belong to a chain.
4. Check the van. Traffic Flow, run at your delivery hour, shows whether the street outside is moving or jammed; it measures speed, not the number of vehicles.

A worked example in Melbourne

A made-up case. A sourdough baker in Melbourne has a wholesale kitchen and wants a shop. Two units are on offer. The figures are invented, to show the reading.

Made-up figures, for illustration only.

Station Street fits a commuter bakery, but it already has two chain bakeries catching the same people and nowhere to unload the wholesale van. Village Lane has twice the homes, one rival and loading at the kerb. Its busiest hour falls on Saturday morning, which suits a sourdough shop with a weekend queue. The baker takes Village Lane and keeps the weekday wholesale run as the steady income.

Mistakes to avoid

• Judging the street at lunchtime. A bakery's day is early. Busy at 13:00 says little about 07:30.
• Counting only bakeries. Cafes with pastries and supermarkets with a bake-off counter take the same breakfast money.
• Forgetting the oven. A deck oven needs power, extraction and floor strength. Rule out units that cannot take one before comparing anything else.
• Reading footfall as customers. Footfall is people-hours of presence around the place, not people who would buy. A hospital car park is busy and sells little bread.

Doing it with an AI employee

An AI employee can check every unit an agent sends against the trade you have chosen. The blocks, in order:

1. My data: candidate units with their addresses.
2. Travel area: a 10-minute walk around each.
3. Nearby places: bakeries, cafes and supermarkets inside the walk, with chain or independent.
4. Demographics: residents inside the walk, where there are figures.
5. Mobility Footfall: footfall per day, busiest hour and the morning share, where the city is covered.
6. xMap LLM: score each unit for your lead trade (commuter, neighbourhood or wholesale) with a reason, from these figures only.
7. Make a report: one page per unit for the landlord meeting or the bank.

The report lands in your inbox with the units in order and their figures beside them. For the wholesale side, Cafes and hotels for my bakery's wholesale finds the accounts to supply, and Where to open my next cafe is a close cousin for the shop itself.