Must be decent at everything: the geometric mean, explained simply
The usual score lets a strength cover for a weakness. Sometimes that is right; for site selection it often is not. 'Must be decent at everything' (the geometric mean, the way the UN Human Development Index works) scores a hex by its weakest link. Here is what changes on a London map.
Give a score two factors of equal importance: demand and accessibility, and take two hexes - a hex is one hexagon on the map, about 350 m across (roughly 0.1 km²). Hex A has the best demand in the city and no transit at all. Hex B is decent at both. The usual score adds the two up - and gives A and B exactly the same number.
For open-ended exploring that is fine: a strength makes up for a weakness. For choosing a site it is usually wrong, because a hex nobody can reach is not half as good as a reachable one - it is out. Must be decent at everything is the other way to combine factors. It multiplies instead of adding, so a hex is only ever as good as its weakest factor. The field calls it the geometric mean, and it is how the UN Human Development Index has worked since 2010 - deliberately, so a strong economy can no longer hide a failing school system.
Must be decent at everythingthe field's term: Geometric mean
Cafés within a 10-minute walk, residents and transit stops within a 10-minute walk, combined by 'Must be decent at everything'. Bright means solid on all three; deep purple means at least one is weak, however strong the others.
The same map, added up instead
Here is the identical set of factors under Balanced, the default, which adds them up with their importance. Compare the two: the balanced map is generous, fading gradually from the bright center through greens and teals; the weakest-link map is decisive, keeping the hexes that pass on every count and dropping the rest.
Balanced: strengths make up for weaknesses, so the outer boroughs fade gradually through greens and teals.Must be decent at everything: only the hexes solid on all three factors stay bright; the rest drop to the bottom.
Switching it on
It is one card at the bottom of the Score map's Add measures pop-up, after the factors it acts on. The description under the dropdown says the consequence out loud, because you should know it before you choose it.
Where to click
Map type
Score
Add measures
How the factors combine
Must be decent at everything
The control, with its consequence in plain words: 'One with none of any one of them scores zero, however strong the rest are.'The default, Balanced, with its own description. Flip between the two and the whole map re-scores.
There is also a ready-made question built on it. Open Questions & guides and look for the one that asks where competition is thin, footfall high and people many, with no weakness anywhere. It opens a Score with people nearby, fewer food places, walkable streets and transit, already combined by Must be decent at everything. The Score's Show the math card names the choice too, in its Combined: line.
One hex, two verdicts - and a hex with a gap
Start with a hex that is decent at everything. Its hex card - what hovering a hex shows - says why: transit stops in the top 1% of hexes (40 within a 10-minute walk), cafés in the top 7% (22 within a 10-minute walk) and residents around the middle. Under either way of combining, it scores almost the same.
Balanced: 89 out of 100. Under Balanced the card gives each factor's points - 33 + 32 + 24 - and they add up to the Score.Must be decent at everything: 88 out of 100. A hex solid on all three hardly notices the switch. The factors are multiplied here, so the card leaves out points.
Now read each factor in just its own hex, and look for a gap. This hex is in the top 2% of hexes for cafés and the top 1% for residents - and has no transit stop (it is level with 96% of hexes, which have none either).
Must be decent at everything: 0 out of 100. Two factors near the top of London, one at zero - and the zero wins. Under Balanced, its two strong factors, each worth about a third, would keep roughly two thirds of the points.
Balanced is just as clear the other way. The next hex, also read in its own hex, has no café and no transit stop, and only middling residents - yet it still scores 24 out of 100, every point from the residents. Under Must be decent at everything, the same hex would score 0.
Balanced: 24 out of 100 for a hex with no café and no transit stop. The residents alone carry it.
Zero is deliberate, and the control tells you
Multiplying by a factor measured at zero gives zero. The app does not soften that: if you say every factor is a must-have, a hex with none of one of them is out. If that feels too strict for a factor - transit within a 10-minute walk is usually the fairer question than transit in one small hex - change that factor's Look at setting rather than the way the factors combine.
With a Range: a zero at the far edge
A factor can also carry a Range - a span of values you want, such as the top 30% of hexes by residents. Set it to Give them fewer points and the factor gives full points inside the span, falling in a straight line to zero at the lowest and highest values among the hexes compared.
Under Balanced, a hex at that far edge just loses that factor's points. Under Must be decent at everything it gets the same treatment as any other zero: the whole Score floors to zero. That is deliberate - a hex as far from your range as a hex can be is not decent at that factor - and it is worth knowing before you combine a range with this setting. A hex merely near the range loses only a little, here as under Balanced.
The math, honestly
Each factor is first placed on a 0–100 scale (see how each factor is placed). Balanced then takes the weighted sum; Must be decent at everything takes the weighted product - equivalently, the average of the logs, which is why a zero is fatal: the log of zero does not exist, so the app treats a measured zero as a fail rather than as missing data.
Two hexes, two factors of equal importance
Hex A - demand 100, accessibility 0, Balanced
(100 + 0) ÷ 2 = 50
Hex A - Must be decent at everything
√(100 × 0) = 0
Hex B - demand 50, accessibility 50, Balanced
(50 + 50) ÷ 2 = 50
Hex B - Must be decent at everything
√(50 × 50) = 50
Balanced cannot tell A from B. The geometric mean can, and it prefers the hex that is decent at both - which is nearly always what site selection wants.
Use Balanced when factors trade off
A bit less footfall for a lot less competition is a real trade, and adding up is the honest way to express it. Also the right default for exploring.
Use Must be decent at everything for must-haves
When each factor is a requirement rather than a preference, multiply. Excellent-on-one, poor-on-another hexes drop; solid-on-all hexes rise.
Balanced adds the factors up, so a strength covers a weakness. Must be decent at everything multiplies them, so the weakest factor rules.
A factor measured at zero floors the score to zero - deliberately, and the control says so before you pick it.
It is the UN Human Development Index method: one strong score can no longer hide a failing one.
The map turns from generous and gradual to decisive: bright islands, deep-purple gaps.
A hex decent at everything scores almost the same either way (89 against 88); a hex with one zero drops to 0.
A Range set to Give them fewer points reaches zero at the far edge - and here that zero floors the Score too.
Quick questions
What is the difference between 'Balanced' and 'Must be decent at everything'?
'Balanced' adds the factors up with their weights, so a strength can make up for a weakness. 'Must be decent at everything' multiplies them instead, so a hex is only as good as its weakest factor.
Why does a hex score zero?
Because it has none of one of the factors - or sits at the far edge of a Range set to Give them fewer points. Multiplying by zero gives zero, deliberately: if there is no transit at all, no amount of demand rescues the hex. The control says so before you choose it.
Is this the same as the Human Development Index?
Yes in method. The UN's HDI switched to a geometric mean so that a strong economy could no longer hide failing health or education. The same logic applies to a location.
When should I use 'Balanced'?
For open-ended exploring, or when your factors genuinely trade off - a bit less footfall for a lot less competition, say. Use the weakest-link option when every factor is a must-have.
Does it change which neighborhoods rank highest?
Often, yes. Hexes that are excellent on one factor and poor on another drop; hexes that are solid on everything rise. Flip between the two and watch which hexes move - that is the story.
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.
