How every score is computed.
We publish a 0–100 score for liveability, safety, transport, schools and rental affordability for every neighbourhood in England and Wales. Every score is built from named public datasets using formulae we publish in full below — no black boxes, no proprietary "weighting that we can't show you".
Last methodology change: 29 April 2026 · See /changelog for prior revisions.
What "an area" means
We use the UK Office for National Statistics geography hierarchy. Coverage is England + Wales (Scotland and Northern Ireland use separate statistical geographies and aren't in v1).
- Region — 10 statistical regions; thousands of local areas each.
- Council area — 374 local authorities; council tax + most salary/jobs data lives at this level. Officially called a "Local Authority District" or LAD.
- Neighbourhood — 7,264 named neighbourhoods, ~7,500 residents each. The default page on the site. Officially a "Middle Super Output Area" or MSOA.
- Local area — 35,672 sub-neighbourhoods, ~1,500 residents each. Most of our metric data lives here. Officially a "Lower Super Output Area" or LSOA.
- Postcode — Royal Mail postcode unit, ~30 properties on average. We map every postcode to its local area.
The five composite scores
Liveability score
Seven-factor composite with non-linear response curves (extreme unaffordability and danger penalised disproportionately), a synergy bonus for areas that are both affordable and safe, a flood zone deduction, and hard caps for nationally dangerous or severely unaffordable LSOAs.
Inputs: 28% affordability, 22% safety, 18% transport, 12% healthcare, 8% air quality, 7% schools, 5% EPC — each through a non-linear curve, plus synergy bonus and flood deduction
Full liveability score formula →Safety score
Lower crime → higher score. National percentile rank, so a 70/100 area is in the safer 70% of England + Wales. Police.uk for England & Wales; Greater Manchester filled with an IMD-derived proxy.
Inputs: Crimes per 1,000 residents per year, percentile-ranked nationally and inverted (so 100 = safest). Surfaced on each page as a multiplier vs. the England + Wales average — the per-resident rate inflates in workday/tourist destinations and is shown only as a supporting figure.
Full safety score formula →Transport score
Combines walk-to-rail (40%), walk-to-metro (20%) and public-transport job access (40%). National rail station file plus TfL + OSM for regional metros, plus DfT JTS commute matrices.
Inputs: Distance to nearest rail station, distance to nearest metro, public-transport time to 5,000+ jobs
Full transport score formula →Schools score
Spatial fan-out: every local-area centroid is matched to schools within a 2km radius via PostGIS, then weighted by Ofsted rating. Means rural areas with one nearby Outstanding school still score well.
Inputs: % of schools within 2km rated Good or Outstanding, distance to nearest Outstanding school
Full schools score formula →Rental affordability
Lower rent-to-takehome ratio = higher score. Caps at 30% (national affordability rule of thumb) and 60% (severely unaffordable). Salary from ONS ASHE, rent from ONS PIPR.
Inputs: Median rent ÷ median post-tax salary at council-area level
Full rental affordability formula →How we roll metrics up to neighbourhoods, cities and regions
A neighbourhood page combines data from its constituent local areas. A council-area page combines all of its local areas. We use population-weighted means for rates and percentages (so a 10,000-person local area influences the average more than a 500-person one) and sums for counts (jobs, dwellings, EPC certificates).
Some metrics are only published at a coarser-than-local-area level — for example, council tax and ASHE salaries are published at council-area level, and demographics from Census 2021 are mostly at neighbourhood level. We explicitly mark these as "council-area estimate" or "neighbourhood estimate" on the page so users know the figure is the council's value applied to every neighbourhood in that council, not a per-postcode truth.
What we're honest about
- Greater Manchester crime — data.police.uk doesn't carry GMP records after 2020. We fill the 1,294 affected GM neighbourhoods with an IMD-derived proxy (national percentile rank inversion of the 2019 Index of Multiple Deprivation Crime Score) and label it as such on every page where it appears.
- Wales metrics — Welsh fuel poverty, council tax bands and tenure mix come from the Welsh Government rather than DESNZ / DLUHC. Some Census tables are England-only; we use the StatsWales equivalent where available, otherwise show "not published in Wales".
- Rent estimates — ONS PIPR publishes rent at council-area level. We multiply this by a local-area "quality multiplier" from our rent-estimate model to give a per-local-area figure. Estimated, not measured.
- Mobile coverage — Ofcom Connected Nations is published at council-area level only. The indoor-any-network percentage is >99% across most of the UK so it's a poor differentiator below city level.