How cloud kitchens pick delivery zones riders can reach in time

A cloud kitchen picks its delivery zone by starting from the clock, not the map: draw everything a rider can reach from the kitchen within the time your food stays good, then look inside that shape for the streets where many people live or work and few restaurants already cook what you plan to sell. The zone is the overlap of reach, demand and a gap in supply. Anything outside it is a delivery that arrives cold, however many orders it might bring.

This is written for operators of delivery-only kitchens, virtual brands run from existing restaurants, and the delivery platforms and landlords who help them choose a unit. It covers the three layers of a zone map, a made-up comparison of two kitchens, and an AI employee that redraws the picture every month.

A zone is a promise about time

Customers judge a delivery by how long it took and what state the food was in. Both depend on distance by road and on the traffic in the evening, not on distance as the crow flies. A kitchen behind a ring road can be 2 km from a dense neighbourhood and still 20 minutes away by motorbike. So the first layer of any zone is a travel area: what can be reached in, say, 10 or 15 minutes by car or by bike from the kitchen door. The Travel area block draws exactly that along the roads, up to 15 minutes or 10 km by road today, by car, bike or on foot.

The three layers of a zone map

1. Reach. The travel area around each candidate kitchen. Draw two sizes, a comfortable one and a stretch one, so you can see what an extra few minutes would add.
2. Demand. Where people are. Demographics can return every hexagon of about 0.1 km² with its residents and age groups, ready for a heat map, in the countries it covers (Thailand, Egypt, India, France, Canada and Australia among them; not the United Kingdom or the United States). Add offices, colleges and hospitals with Nearby places for lunchtime demand.
3. Supply. The restaurants already inside the zone, by cuisine. Nearby places counts them by type, with chain, rating and opening date, so you can see whether the zone is crowded with burgers and short of everything else.

Restaurant markets differ a lot in how chained they are, which changes who you compete with. xMap's places database lists 1,640 restaurants within 10 km of the centre of Sydney (November 2026); of the 303 nearest the centre, only 10 belong to a chain. A virtual brand there competes mostly with independents, many of whom also deliver.

A worked example: two kitchens in Bangkok

Say an operator is choosing between two units in Bangkok, one near a large residential area to the east and one in a commercial district closer to the centre. These figures are made up to show the method. Within 12 minutes by motorbike, the eastern unit reaches about 180,000 residents but only a handful of offices; the central unit reaches about 95,000 residents and several large office towers and a university. Inside the eastern zone, Nearby places finds dozens of Thai restaurants and noodle shops but very few places serving salads, poke or Japanese set meals. The central zone has plenty of all three.

The operator's plan was a healthy-bowl brand. In the central zone it would be one of many. In the eastern zone it fills a gap for a large residential population, with dinner rather than lunch as the main shift. The kitchen moves east, and the menu adds family portions. The numbers decided the unit; they also changed the product.

The zone review as an AI employee

1. Schedule: the first Monday of each month.
2. My data: your kitchens and any candidate units, with positions.
3. Travel area, Each incoming place: 12 minutes by bike or by car around each.
4. Demographics: residents inside each zone, where the country has figures.
5. Nearby places: restaurants inside each zone, with several cuisine types picked so each count comes back separately.
6. Calculated field: residents per restaurant of your cuisine.
7. xMap LLM, on the whole list: which cuisines are thinly served in each zone, using only the counts.
8. Make a report: a PDF with a map of every zone and the gaps, to the operations lead.

For the evening itself, a second flow can check Traffic incidents inside each zone every hour from 17:00 and message the kitchen when a closure cuts the zone in two, so the platform radius can be shortened before orders run late.

Mistakes that lengthen deliveries

• Drawing the zone as a circle. Circles count streets riders cannot reach in time. Use travel time.
• Planning on daytime roads. Roads are slower at dinner time. Draw the comfortable zone smaller than the travel area suggests, or check live speeds with Traffic Flow at the hour you deliver.
• Counting residents only. Office districts empty at night and fill at lunch. Decide which shift you are building for.
• Ignoring restaurants that deliver themselves. Every restaurant in the zone is a possible rival, not only the ones on your platform.
• Assuming population data in every country. Where there is none, rank zones by offices, schools and the restaurants already there, and say so.

Templates to start from: where to open a delivery kitchen, which cuisine is missing in my delivery zone, cuisines missing in each district and road closures in my delivery zones. The food delivery page has more.