People per square kilometre: what counts as dense, and compared with what

There is no single number of people per square kilometre that makes a place dense. The figure depends on how big an area you measure, and what feels dense in one city is ordinary in another. The useful question is "dense compared with what": compare an area with others of the same size in the same city, or with the places where your business already works well. Set your own bands from those, and "dense" starts to mean something you can act on.

This is for anyone who has seen people per km² in a report and wondered whether the number is high: retailers choosing a format, delivery companies sizing zones, planners deciding where a service is stretched. It explains how the figure is made, why it moves with the size of the area, and how to read it.

How the figure is worked out

People per km² is residents divided by area. In xMap Flows, Demographics gives people_per_km2 for each place, area or hexagon it measures, from the residents it counts and the size of the area (area_km2). A hexagon of the grid is about 0.1 km², so a hexagon holding 400 residents works out at about 4,000 people per km². A 1 km circle is about 3.1 km², so 25,000 residents within it is about 8,000 people per km².

Note what it counts: people who live there. An office quarter can have a very low figure and a very crowded pavement.

Why the size of the area changes everything

The bigger the area, the more empty land it takes in: parks, roads, rivers, rail yards, factories. So the same neighbourhood gives very different figures depending on how you draw it.

Invented figures for one made-up neighbourhood, to show how the measure moves with the area.

Notice the 500 m figure is higher than the single hexagon. That is not a mistake: the hexagon picked was a mixed block with shops, while the circle around it is mostly apartments. Small areas swing both ways; large areas drift down as they take in empty land. Comparing a hexagon from one place with a city average from another tells you nothing.

Setting your own bands

Instead of borrowing a threshold, build one from the places you know.

1. Pick a scale that suits the decision. For a shop people walk to, 500 m or 1 km around the door. For a delivery zone, the area the zone covers.
2. Measure people per km² at that scale for a set of places you know well: your own sites, or twenty points spread across the city.
3. Sort them and split the list into three or four equal groups. The top group is "dense" for this city and this scale; the bottom is "loose".
4. Add the result you care about next to each, such as sales or bookings, and see which band your best places fall in. That band is your target.

Bands made this way travel badly between cities. A dense band in a compact European city and one in a sprawling North American or Australian city will sit at different numbers. Build bands per city, or at least per kind of city.

A worked example in Mumbai

This is a made-up example. A grocery delivery start-up runs dark stores in the western suburbs of Mumbai and wants to know which kind of area its best stores serve. It measures people per km² within 2 km of each of its 12 stores. Sorted, the figures fall into three clear groups, and the start-up names them by its own data: "very dense", "dense" and "mixed". Its four stores with the most orders per day are all in the "very dense" group.

When it looks at five candidate sites in a new part of the city, it runs the same 2 km measure and keeps only the two that fall into its own "very dense" band. The number that made a site "very dense" came from its own stores, not from a rule of thumb.

Mistakes to avoid

• Borrowing a number from an article or another country. A threshold that means dense in one place means suburban in another. Build your own.
• Mixing scales. A hexagon's figure and a 2 km circle's figure are different measures of the same place.
• Letting water drag a coastal site down. A circle half over the sea halves the residents and doubles the apparent emptiness. Draw a shape over the land, or look at the hexagons near the site.
• Reading density as demand. A dense area of young renters and a dense area of retirees buy different things. Look at the age groups too.

Doing it with an AI employee

1. Bring your sites in with My data, or put candidate points in an Area of interest.
2. Add Demographics at the distance you chose. Each site gets population, area_km2 and people_per_km2.
3. Add a Calculated field that puts each in your bands, such as band = if(people_per_km2 > 20000, "Very dense", if(people_per_km2 > 12000, "Dense", "Mixed")), with the cut-offs from your own sites.
4. Add Sort and limit, densest first, and end with Email or a Google Sheets app block.

What lands in the inbox is every site with its residents, its density and the band it falls into, in your own words. For a picture rather than a list, a second Demographics block set to Every hexagon feeding Show on a map shows where the density sits street by street.