What street-level demographic data shows, and what it leaves out

Street-level demographic data tells you how many people live in a small patch of a city, how many of them are men and how many are women, and how old they are in ten-year bands. That is all it holds. It does not tell you what those people earn, what they spend, where they work or whether they shop nearby. Used for what it is, a count of residents by age, it answers a surprising number of business questions on its own.

This article is for owners, planners and analysts who have been handed a population map and want to know what they can safely read from it. It walks through each figure, the questions it answers, the questions it cannot, and how an AI employee in xMap Flows puts the figures next to the places you care about.

The figures, one by one

xMap's population grid splits the land into small hexagons of about 0.1 km² each, roughly the size of a few city blocks. For every hexagon it holds the number of people who live there. The Demographics block adds those hexagons up for whatever area you ask about and gives back these fields:

• population: everyone living inside the area.
• population_male and population_female, with male_pct, the share of men.
• Nine age groups: age_0_9, age_10_19 and so on up to age_70_79, then age_80_plus.
• Three broader groups: children_0_19, adults_20_59 and seniors_60_plus, with children_pct and seniors_pct.
• area_km2 and people_per_km2, so a wide suburb and a tight city centre can be compared fairly.

Every one of these is a count of residents: people who sleep there. A hexagon covering an office tower or a shopping centre may show very few people, because hardly anyone lives in it, even though thousands pass through each day.

Questions it answers well

Because the grid is fine, it can tell two neighbouring districts apart, which a city-wide average never can. Questions it handles well include:

• How many people live within 1 km of each of my shops, and which shop has the fewest?
• Which parts of the city have the most children under ten, for a nursery or a paediatric clinic?
• Where is the share of people over 60 highest, for home care or a pharmacy?
• Is this district dense (many people on each km²) or spread out, which changes how far people will walk?
• Roughly how many residents would a new branch reach compared with an existing one?

Each of these is about where people live and how old they are. When the question fits that shape, the answer is direct and easy to explain to a board or a client.

What it does not hold

It is just as important to know the gaps, because a population map looks authoritative and invites reading too much into it.

• No income or wealth. A hexagon with 3,000 residents could be a quiet, well-off suburb or a crowded, low-income one. The grid cannot tell them apart.
• No spending. There is nothing about what people buy, how often, or where.
• No workers or visitors. The people who commute into a business district are not in it. For how busy a place is during the day, you need footfall, which is a different measure (see what footfall data measures).
• No households. The figures count people, not homes or families. Many children in an area suggests families, but the data does not group anyone into households.
• No ethnicity, language, education or jobs.

A worked example

Take a made-up example. A bakery chain in Lyon has three shops and wants to know which one has the most people within a short walk. It asks for everyone within 500 m of each shop. Suppose the answers come back like this (the figures are invented for illustration):

A made-up example to show how the fields read; not real figures for Lyon.

Shop C has the most people around it and the oldest neighbours. Shop B has fewer residents but the largest share of young people, which may explain a busier after-school trade. What the table cannot say is which shop's neighbours have more money to spend, or how many office workers walk past Shop A at lunchtime. Those are questions for other data, or for the shop managers who know their streets.

Getting the figures with an AI employee

You can ask once, or have an AI employee check every month and email you the result. The usual order of blocks is:

1. Bring in your places: My data with a table of your shops, or an Area of interest for whole cities or districts you draw.
2. Add Demographics. Set Where to Each incoming place and pick the distance under Within, from 500 m to 8 km. For whole areas, choose Totals for each area.
3. Add Sort and limit to put the places with the most residents, or the highest share of an age group, first.
4. End with an Email or a map, so the list lands in your inbox ready to read.

The template Demographic profile of each city does this for a set of cities and writes each one's profile into a sheet. Who lives around new places adds the residents within 1 km of every new cafe it finds, so the list starts with the busiest neighbourhoods.

Mistakes to avoid

• Treating residents as customers. A count of people is a ceiling on who might come, not a forecast of sales.
• Comparing a 500 m circle in one city with a 2 km circle in another. Use the same distance everywhere, or compare people_per_km2.
• Reading a 0 as an empty area. It usually means the country is not covered.
• Filling the gaps with guesses. If income matters to the decision, say so in the report rather than reading it into a population map.