Methodology

How a score is built, checked and labelled.

Every number on the map comes from a public source or from a model fitted to them, and says which. Here is the whole chain, including what we don't know.

Talk to us Signals & sources

15,341: pincodes scored
769: with data of their own
District: level of the spending anchor
Weekly: refit, gated

The chain

Six steps from public data to the map

Each step is automated and re-run on a schedule; none is edited by hand.

Collect
Pull each source in its published form (RBI banking, income-tax filers, vehicle registrations, satellite night-lights, UPI value and more), keep its licence terms and record when it was fetched.
Place
Attach every value to a pincode using India Post's PIN boundaries and directory. A pincode's name is its own India Post office name, and its point sits inside its own boundary.
Anchor
Tie household spending to the national survey: Household Consumption Expenditure Survey 2023-24 figures at district level, from a compilation of MoSPI's unit-level data (under review: MoSPI publishes official figures only by state). 100 on the purchasing-power index is the district norm.
Model
Three models (a regression on the combined signals, a gradient-boosted model and a spatial smoother) are blended into purchasing power; income and spend follow from the anchor.
Estimate gaps honestly
A pincode with no data of its own takes its district's typical score and is labelled "District estimate". We don't invent differences we can't measure.
Check before publishing
The weekly refit must keep fixed orderings (for example Golf Links above Saket in Delhi) and may not swing pincodes by more than 10 points without a dated, approved review. Data sanity checks run every night.

Coverage, plainly

What rests on data, and what is estimated

15,341: pincodes scored80% of India Post's 19,097
769: have data of their ownabout 5% of scored pincodes
Labelled: every other scoreas a district estimate

Counts are read from the data each time this site is built.

Reading the map

The labels you'll see

Own data: The pincode has at least one measured value of its own; its score can differ from its neighbours'.
District estimate: No data of its own: the typical score of its district. Two such pincodes in a district are deliberately equal.
City-level estimate: Property rate. Indicative for the city, not a valuation or transaction price.
—: No value. We show a blank rather than fill a gap.

Our rules

What we always do, and never do

Always

Say where a number comes from and at what level
Show an estimate as an estimate
Report a failed check rather than work around it
Quote coverage as measured pincodes next to the total

Never

Invent places, numbers or sample data to fill a gap
Let a district or state figure separate two pincodes
Loosen or swap a check to get a refit through
Claim footfall, transaction or real-time data

FAQ

Methodology questions

Are these real incomes?

No. They are modelled estimates of a typical household, good for comparing areas, not for judging any one household.

Why do neighbouring pincodes sometimes show the same score?

Because neither has data of its own, so both take their district's estimate. Showing a difference we can't measure would be invented precision.

Why did a score change?

The model is refitted weekly as sources update. Large changes are only allowed through a dated, reviewed approval.

Can we see the sources?

Yes. The Signals & data page lists every signal with its source and level.

Want to check a score with us?

Send us a pincode and the decision behind it; we'll show you exactly what the score rests on.

Talk to us Open the free map

How a score is built, checked and labelled.

Six steps from public data to the map

Collect

Place

Anchor

Model

Estimate gaps honestly

Check before publishing