Per 1 000 residents: turning a count into a rate, explained simply
Two hexes both have one pharmacy. One has 300 residents, the other 6 000. A plain count says they are the same. 'Per 1 000 residents' says the first is twenty times better served. Here is the formula, and the small-numbers trap it has to avoid.
A hex is one hexagon on the map, about 350 m across (roughly 0.1 km²). One of these two has 300 residents. The other has 6 000. A plain count - one pharmacy - says they are identical. Anyone queuing for a prescription knows they are not: the first is twenty times better served.
Per 1 000 residents is the one-step fix. Divide the count by the people who live there, scale it up to a thousand so the number is readable, and you have a rate: 3.3 pharmacies per 1 000 residents in the first hex, 0.17 in the second. It is the most ordinary formula on the map and the one that changes the most maps, because nearly every “where is there a lot of X” question is really “where is there a lot of X for the people there”.
Per 1 000 residentsthe field's term: Rate
Pharmacies per 1 000 residents, read within a 10-minute walk and placed by rank in a Score so the whole spread of color is used: brighter hexes have more pharmacies for the people who live around them.
Switching it on
The map for “how much is here” is called Amount, and its Count as row is where a count becomes a rate. The same row sits on each factor of a Score. Its default is “Total” - the count as it is. Change it to “Per 1 000 residents” and the number, the legend and the badge all switch to the rate.
Where to click
Map type
Amount
Pharmacy
Count as: Per 1 000 residents
The badge reads 'Per 1 000 residents · Rate' and the sentence under it, 'Pharmacy per 1 000 residents, within a 10-minute walk', is exactly what the map draws.'Count as' offers Total, Per 1 000 residents, Per km² and Share of its category - plus the census bases below, where the census publishes them. Options a measure cannot support are grayed out with a reason.
Here the pharmacies are read within a 10-minute walk: the places you can reach on foot within 10 minutes, along streets and paths, nearer ones counting more (full weight under 3.5 min, 70% to 7 min, 40% to 10 min). That count is divided by the people living around the hex - in the hex and the hexes around it - so the rate reads as “pharmacies within reach, for the people who live here”. The walk post goes through it.
Why 1 000 and not 1?
Because 0.0033 pharmacies per person is unreadable. Per thousand gives 3.3, a number people can hold in their head and compare across hexes.
Why is it grayed out for some measures?
An average income or a share is already a per-something number. Dividing it by residents again is meaningless, so the app refuses and says why.
Reading the map
On the Amount map the legend is banded: one color per band, from the smallest rate to the largest, and a line saying how the bands were chosen. Hover any hex and the card gives its rate.
Banded, by natural breaks. The note is honest about London: 59 in every 100 hexes have no pharmacy within a 10-minute walk, so the lowest bands start at the same number.One hex in south-east London: 0.2 pharmacies per 1 000 residents within a 10-minute walk.
In a Score the same rate is placed by rank, and hovering a hex opens the hex card: the hex's score, the points each factor earns, where it stands among London's hexes, and a warning when what you asked for is rare. Click a hex and Hex details holds the same card, with ‘Show the math’ for the division.
The hex card on the Score: 90 of 100 points for pharmacies per 1 000 residents, in the top 11% of hexes - and a reminder that 6 in 10 London hexes have none within a 10-minute walk.
The math, honestly
Pharmacies per 1 000 residents in one hex (illustrative)
Pharmacies counted
2
Residents
6 000
Divide
2 ÷ 6 000 = 0.00033
Per 1 000 residents
0.00033 × 1 000 = 0.33
The 300-resident hex with one pharmacy gives 1 ÷ 300 × 1 000 = 3.3. Ten times the rate, from the same 'one pharmacy'. Residents come from ONS Census 2021 and are an estimate for each hex.
The trap: tiny hexes that shout
Here is what would happen without a safeguard. A hex on the edge of a park has 3 residents and 1 pharmacy. The rate is 1 ÷ 3 × 1 000 = 333 per 1 000. It would be the brightest hex in London, and it would be meaningless - three people prove nothing.
MapBees uses the standard fix from disease mapping, called empirical Bayes smoothing. The rule in one sentence: when a hex has lots of residents, trust its own rate; when it has almost none, pull it toward the city's rate. The 3-resident hex collapses to something near London's average; the 6 000-resident hex barely moves, because it has real evidence.
The adjustment is declared, not hidden: 'where the base is tiny, the value is pulled toward the city's, so a handful of people cannot produce an extreme rate'. The method card lists it among the formulas.
What 'adjusted for small numbers' means on a hover card
It means the hex's own rate was more than half replaced by the city's. Treat the number as “probably about average, we cannot tell” rather than as a measurement. Hexes with nobody living in them show no rate at all - a pharmacy per resident in an empty park is not zero, it is a question that does not apply.
Other bases: children, residents 65 and over, building volume
Residents are not always the right people to count against. A playground serves children; a pharmacy matters most to older residents. Where the census publishes the figures, Count as offers more bases:
Per 1 000 children under 10 - the field's term in the badge is “Rate per child”;
Per 1 000 residents 65 and over - “Rate per senior”;
Per 100 000 m³ of residential building volume - “Rate per building volume”;
Per 1 000 households - but not in London. MapBees has no household figures for London, so the option is grayed out with its reason: “There are no figures for households in London, so nothing can be counted per households here.”
Where to click
Map type
Amount
Playground
Count as: Per 1 000 children under 10
Count as for playgrounds. Per 1 000 households is grayed out in London; the other census bases are there.Playgrounds per 1 000 children under 10, in the hex itself. Fewer hexes are measured than for residents: a rate per child needs children to count against.
Playgrounds per 1 000 children under 10. Most of London sits in the darkest band; the bright hexes are single hexes with a playground and few children. Gray means no data, or a rate per child does not apply.
The legend's note warns that the darkest band covers a 45-fold range, so most hexes look alike on this map.A hex in south-east London: 0.1 playgrounds per 1 000 children under 10.
The arithmetic is the same division with a different base, and the same small-numbers adjustment applies: a hex with two children and a playground is pulled toward the city's rate, exactly like a hex with three residents.
Rate versus density
“Per 1 000 residents” answers how well the people are served. Per km² answers how packed the ground is. Use the first for anything people need near home - pharmacies, schools, grocery stores; use the second for anything about footfall and competition on a street. And once you have a rate, the natural next question is whether it is high or low for London, which is the Fair share post.
Per 1 000 residents = count ÷ residents × 1 000. A count becomes a rate you can compare across hexes with any number of residents.
One pharmacy for 300 people and one for 6 000 are not the same. The rate says so; the count cannot.
Tiny bases are pulled toward the city's rate and flagged as adjusted, so three residents cannot light up the map.
Count as also offers children under 10, residents 65 and over and residential building volume where the census publishes them; London has no household figures.
The Amount map shows the rate in bands; in a Score the rate is placed by rank, which spreads the color.
Hover for the rate, or the hex card on a Score; 'Show the math' writes the division out.
Quick questions
What does 'Per 1 000 residents' do?
It divides a count by the number of people living in the hex and scales it to a thousand, so 14 cafés among 1 660 residents becomes 8.4 cafés per 1 000 residents - a number you can compare across hexes with very different populations.
Why per 1 000 and not per person?
Because 0.0084 cafés per person is unreadable. Scaling to a thousand gives numbers like 8.4, which people can hold in their head and compare.
What happens in a hex with almost nobody living in it?
One ATM among three residents would be 333 per 1 000 - nonsense that would dominate the map. MapBees adjusts for small numbers: where the base is tiny, the value is pulled toward the city's rate, and the hex is flagged as adjusted.
Why is 'Per 1 000 residents' grayed out for some measures?
Because some measures are already rates (an average income, a share). Dividing a rate by residents again gives nonsense, so the option is disabled with a reason.
Can I count per child or per older resident instead?
Yes, where the census publishes the figures: Count as also offers Per 1 000 children under 10, Per 1 000 residents 65 and over and Per 100 000 m³ of residential building volume. London has no household figures, so Per 1 000 households is grayed out there with its reason.
What does 'Rate' under the badge mean?
It is the field's own term for a count divided by a population. MapBees leads with the plain words, 'Per 1 000 residents', and shows the term beside them for anyone who wants it.
See it on a real map
The best way to understand the score is to play with it. The full app is on a Free Tier, with no sales call.
Builds MapBees's location-intelligence pipeline end to end — census and Overture Maps ingestion, H3 hex scoring, and the map interface — and writes the methodology behind every number on the site.
