FlowsBook a callSign inCreate free accountSign up
Population and demographics

Using population to split one delivery zone into fair parts

When a delivery zone grows too big for one team, split it by the people who live there, not by area on the map. A practical method with population hexagons and drawn zones.

Part of the guide Location intelligence automation

By the xMap team · · 5 min read

To split a delivery zone fairly, divide it so that each part holds about the same number of residents and can still be reached quickly from the place it is served from. Splitting by area on the map gives one team a dense block of flats and another a stretch of parks and warehouses. Splitting by population gives each team a similar share of likely orders, and the travel check keeps the parts practical.

This is for operations leads at grocery and meal delivery services, laundries, pharmacies that deliver, and any business whose riders or vans serve one zone that has outgrown one team. It covers what population data can and cannot tell you, a step-by-step method, a made-up split in Lyon, and how to keep the split up to date.

What population tells you about orders, and what it does not

Residents are the best simple stand-in for home orders: groceries, dinner, prescriptions and dry cleaning mostly go to where people live. xMap's population grid gives the residents of small hexagons of about 0.1 km² each, with men and women and nine age groups, so you can see where people are concentrated block by block.

Three limits matter:

  • It counts residents, not workers. A lunch delivery service that serves offices needs a different measure; Mobility Footfall shows how busy an area is at midday.
  • It is not available everywhere. It covers about 100 countries, among them France, Canada, India, Thailand, Egypt and Australia. It does not cover the United Kingdom or the United States, so zones there need another basis, such as your own order history.
  • It is not spending. Two districts with the same residents can order very differently. Population sets a fair starting line; your own order data refines it.

The method, step by step

  1. Map where people live. In an Area of interest, draw your current zone. Demographics set to give back every hexagon, then Show on a map as a heat map, shows the dense and empty parts at a glance.
  2. Sketch the parts. Draw two, three or four shapes over the zone, following roads, rivers and rail lines that already divide it, so a rider never has to cross a barrier mid-route.
  3. Count each part. Run Demographics again with the drawn parts and give back totals for each area: one line per part with its residents and age groups.
  4. Check the travel. Travel area from each hub or kitchen, for example 12 minutes by bike, shows whether each part can be reached in the time you promise.
  5. Move the lines and count again. Shift a boundary a few streets until the parts are within a tenth of each other, and every part sits inside its hub's travel area.
  6. Keep the result. Save the final shapes; the next time you change the split, you start from them.

A made-up split in Lyon

The figures in this section are invented to show the method. A grocery delivery service runs one hub in Lyon, and its zone now holds about 210,000 residents, too many for one team of riders on a Saturday. The heat map shows a dense band along the two rivers and quieter streets on the hills to the west. A first sketch cuts the zone in three equal slices on the map: the counts come back at 104,000, 71,000 and 35,000. The western slice is mostly hillside.

The operations lead moves the lines. The dense riverside band is split between two parts at a main bridge, and the western part is pushed further into the centre to pick up a few busy streets. The second count comes back at 72,000, 69,000 and 68,000. A travel check from the hub at 12 minutes by bike covers all of the first two parts and most of the third; the uncovered corner is served from a second, smaller pickup point. Each team now has a similar number of residents in reach and no rider crosses a river twice in a shift.

Use the age groups too

Totals hide who lives there. A part with many residents over 60 may order medicines and groceries more than takeaway at midnight; a part full of people in their twenties may do the reverse. Demographics gives nine age groups and the shares of children and seniors for every part, so two parts with the same total can still be planned differently, for example with different delivery windows or a different product range promoted in each.

Mistakes to avoid

  • Splitting along postcode lines. Postcodes were drawn for mail, not riders; they cut across main roads and lump very different streets together.
  • Cutting through dense blocks. A boundary drawn through the middle of a block of flats moves hundreds of people with a small nudge of the line, and the counts jump about. Draw boundaries along streets.
  • Splitting once and forgetting. New housing changes the balance. Recount the parts each quarter with a scheduled flow.
  • Using residents for office trade. If most orders go to workplaces, population will mislead you.

A scheduled flow can recount the parts each quarter and email the totals, so the operations lead sees when one part starts to outgrow the others. The Same-day pet food delivery zones around my stores and Who my warehouse reaches in 15 minutes templates show how travel areas and population fit together.

Questions

How many residents should one delivery team cover?

There is no general number; it depends on how often people order and how long each drop takes. Start from your own orders per resident and the drops a team can make in a shift, then size the parts to match.

Can I do this in London or New York?

Not with xMap's population grid, which does not cover the United Kingdom or the United States. Use your own order history, or the number of homes and businesses in each part, as the measure instead.

How accurate are the hexagon counts?

They are modelled estimates of residents for small areas, good for comparing parts of a city. Do not treat a single hexagon's number as exact; totals over many hexagons are more reliable.

Does it help with where to put a new hub?

Yes. Count the residents within a travel time of each candidate spot and compare. The part of the zone that is served last and slowest is often where the next hub belongs.

Templates that do this

Read next

More on population and demographics · Every article · Templates

Put it to work

An AI employee in xMap Flows does this on live data, on its own timetable. Start from a template or describe it in plain words.