Population and delivery demand: sizing a zone before you open a kitchen or dark store

Delivery demand is capped by the number of people living inside the area your couriers can reach within the time you promise. Count the residents inside that reach, not inside a round circle, and you have the ceiling for a kitchen, a dark store or a same-day hub. Then three things decide how close to the ceiling you get: how densely those people live (which sets how many drops a courier makes an hour), who they are (age and household), and how well the area is already served.

This article is for delivery platforms, dark-store operators, restaurant groups opening delivery-only kitchens and online shops planning local fulfilment. It explains how to measure the reachable population, what to put next to it, a worked comparison of three sites in Cairo, the usual errors, and an AI employee that keeps the numbers fresh.

Start with the reach, not a radius

A promise like "in 20 minutes" means a courier must get there in perhaps 12 to 15 minutes of riding after the order is packed. The area that covers depends on the roads, rivers, bridges and one-way systems around your door. The Travel area block draws it along the roads, by car, bike or on foot, up to 15 minutes or 10 km by road for now. It uses usual traffic, not the traffic of the moment.

A round 3 km circle drawn on the same site can include a district on the far side of a river with one bridge, which in practice is 25 minutes away. Counting its residents would flatter the site.

Count who lives inside it

Put Demographics after Travel area and it gives the residents inside each reach, men and women, nine age groups and people per km². For most food and grocery delivery, adults aged 20 to 59 are the core; for nappies and baby products, households with young children; for pharmacy delivery, the over-60s may matter most. Population figures exist only in some countries, among them France, Canada, India, Thailand, Egypt and Australia, but not the United Kingdom or the United States.

What else drives orders

• Density. The same number of residents packed into half the area means shorter trips and more drops per courier hour. people_per_km2 is the figure to watch.
• Supply already there. Count the restaurants, supermarkets or pharmacies inside the reach with Nearby places. Many shops nearby can mean people walk out for what they need; few can mean an unmet need, or simply an area with little spending.
• Daytime crowds. Office districts order lunch for people who do not live there. Residents miss them; Mobility Footfall shows how busy the area is at midday, in people-hours, with about 10 days' delay.
• Weather. Rain and heat push orders up on the day. The Weather forecast block helps plan staffing, not site choice.

A worked comparison in Cairo

A made-up case. A grocery delivery company is choosing between three units for a dark store in Cairo, promising delivery within 20 minutes. For each, it draws a 12-minute cycling area and counts what is inside. Every figure here is invented, to show the comparison.

Made-up figures, for illustration only.

Unit 3 has the most people and would win a ranking by residents alone. But it also has the most supermarkets inside its reach, so many of its residents can walk to a shop in minutes. Unit 1 has fewer people but packed tighter, which means faster drops, and fewer competing shops per resident. Unit 2 has the least supply, which looks like an opening, but its residents live far apart, so each courier delivers fewer orders an hour.

A simple way to weigh them is residents per supermarket in the reach: about 8,600 for Unit 1, 20,000 for Unit 2 and 6,900 for Unit 3. The company would view Units 1 and 2 first, then test demand with a short trial on its existing app before signing a lease. Data narrows the list; trial orders decide.

Mistakes to avoid

• Taking residents as orders. They are the ceiling, not the forecast; only your own order history converts one into the other.
• Drawing the reach by car when couriers ride bikes or motorbikes, or the other way round.
• Measuring at night-time population only, for a lunchtime business in an office district.
• Forgetting the second-best site. If the first unit falls through, the comparison table should already show the next.
• Comparing countries with and without population data in one list.

An AI employee for zone sizing

You can run the comparison once for a decision, or monthly to watch your live zones as supply around them changes.

1. My data: candidate units, or your current stores.
2. Travel area: 12 minutes by bike from each.
3. Demographics: residents, age groups and density inside each reach.
4. Nearby places: supermarkets, or restaurants for a kitchen, counted inside each reach.
5. Calculated field: residents per competing shop; then xMap LLM to rank the units and explain the ranking in two sentences, using only the figures given.
6. Make a report: a short PDF to the expansion team.

The report lands with one page per unit and a ranked summary. Ready-made starting points are Where to put the next dark store, Where to open a delivery kitchen and Who my warehouse reaches in 15 minutes, for same-day delivery; for a whole new city, Is a new city ready for delivery?.