What the Representation Score actually measures

Every dataset looks confident until you try to use it. A demographic profile can be perfectly accurate for a whole city and still tell you almost nothing about the four blocks you care about. The Representation Score exists to surface that gap before it becomes a decision.

This article explains what the score measures, how we calculate it, and — just as importantly — what it does not tell you.

The problem it solves

Population statistics are collected at a fixed geography. When you draw a radius, a suburb, or a custom catchments area, you are slicing those fixed geographies with an arbitrary boundary. A single SA1 can straddle the edge of your area, and the people inside it are counted as all-in or all-out.

• Fixed collection geographies rarely line up with the area you actually care about.
• Boundary effects get worse as your area gets smaller.
• A high total population can hide a thin sample underneath it.

How the score is calculated

The Representation Score combines three signals into a single 0–100 number. Each signal answers a different question about how much you should trust the profile for that area.

1. Coverage — what share of the area's estimated population is observed directly rather than modelled or imputed.
1. Granularity — how fine the underlying collection geographies are relative to the size of your area.
1. Stability — how much the profile would move if a single small geography changed.

Image with heatmap panel open

Reading your score

Treat the score as a confidence dial, not a quality grade. A low score does not mean the data is wrong; it means the area is a poor fit for that particular dataset, and you should widen the area or switch to a coarser profile.

Image focusing on heatmapped area

Improving representativeness

• Prefer larger areas when the score is low and you only need directional insight.
• Cross-check a low-scoring area against a neighbouring one to see if the trend holds.
• Use the score alongside other signals — never as the only input to a decision.

Used honestly, the Representation Score turns an invisible source of error into something you can see and argue about. That is the whole point.