How we score councils

Methodology v2.0 (stable). Service quality across seven services (adult social care now includes CQC's assessment), value for money on four (spend compared with councils of similar need and results), and financial health (reserves, debt costs and borrowing, schools-funding deficit, government bailouts).

The measures and weights are fixed. Any change to them will come as a new, dated version, listed in the changelog. The figures still update when the official sources publish new data. If you think something is wrong, the corrections page explains how to tell us and how we respond.

Who is ranked

The 32 London boroughs, 36 metropolitan boroughs and 62 unitary councils: 130 councils in all. These are England's single-tier councils: each runs almost all local services itself, so they can be compared fairly on the same measures. In the rest of England those services are split between a county council and a district council, so neither could be scored on the full set, and we don't rank them. Services where both scores and data are published are weighted by how much they matter to residents and budgets: adult social care and children's services count one and a half times as much as other services. The Isles of Scilly is shown but not ranked: with about 2,300 residents, its per-resident figures are not comparable with other councils. The City of London is shown but not ranked: with about 15,600 residents, its per-resident figures are not comparable with the boroughs. Each council is ranked against all 130, and separately within its own group (London boroughs, metropolitan boroughs or unitary councils).

Separate leagues

Each country is ranked in its own league: England, Scotland and Wales. The services councils run, the bodies that inspect them and the way each measure is defined all differ between the countries, so a rank in one league cannot be compared with a rank in another, and we do not combine them into one table. The method is the same, with the same pillars and the same way of combining scores; the measures are not. England is stable; Scotland and Wales are provisional. Each league has its own methodology page, linked from its home page.

What we measure

How the scores are combined

1. Each measure is converted so that higher always means better.
2. Extreme values are capped at the 5th and 95th percentiles, so one outlier can't dominate. The exception is the CQC assessment score for adult social care: it is a bounded scale set by the regulator, and its extremes (an Inadequate or Outstanding rating) are exactly what it is meant to show.
3. Each measure becomes a standard score: how far a council is above or below the average, in standard deviations.
4. Service-area scores are weighted averages of their measures. A council must have data covering at least 60% of the weight to receive a score. If a measure is missing, the council's other measures in that area share its weight in proportion to their own; the same applies when a whole service area or pillar is missing. Councils scored on incomplete data are marked with an asterisk in the rankings.
5. Service areas are combined into pillars, and pillars into the overall score, using the weights shown above. Scores are re-standardised at each step, so the published weights are the weights that actually apply.

Value-for-money spending models

Expected spend comes from a least-squares regression across all 130 councils (separate fits are used within each group). Each figure is the change in expected spend for a one-unit change in that factor. R² is the share of the differences in spending between councils that the model explains; where it is low, need and results explain little of why councils spend different amounts, and the value-for-money score is close to simple spending per head. Treat those scores with more caution.

Service quality, adjusted for need

Alongside the main ranking we show a second view of service quality that allows for how much need each council's area has. It answers a different question: not "how well does this council do?" but "how well does it do compared with councils whose areas have similar needs?" It is not part of the overall rank, and we don't publish an adjusted overall rank.

For children's services, adult social care and waste & recycling, we measured how each service's score tends to move with its need measure across the 130 councils, and took that part out. What is left is the adjusted score. Planning, complaints, special educational needs and homelessness are left as they are, because there is no published measure of their need. We use the same need measures as the value-for-money models, except that we leave out the London cost marker: London boroughs, metropolitan boroughs and unitary councils already have their own views, and allowing for London would hide real differences between them.

Effects are in standard deviations of the service score, per unit of the need measure.

What this shows. Deprivation explains little of the variation in service quality across these councils: the rank correlation between service quality and the Index of Multiple Deprivation is −0.09. So allowing for need changes the order only a little. The adjusted and unadjusted ranks have a rank correlation of 0.98, the largest move is 20 places, and 58 of 130 councils move more than 5. Treat moves of a few places as no change.

How sure are we? The adjusted rank has a likely range, worked out like the overall one (see below), and each re-run also refits the need models on a random resample of the councils. The ranges are wide. For 127 of 130 councils the unadjusted rank sits inside the adjusted range, so for most the difference between the two views is smaller than the uncertainty in either.

For 3 councils the difference is larger than the uncertainty: Manchester (22nd unadjusted, 7th adjusted, likely range 4th–21st), Westmorland and Furness (12th unadjusted, 21st adjusted, likely range 13th–34th) and Central Bedfordshire (47th unadjusted, 63rd adjusted, likely range 48th–78th). Allowing for need moves councils whose areas are younger and more deprived up, and those whose areas are older and less deprived down, because people in older, better-off areas tend to report more satisfaction with social care and those areas recycle more. We don't adjust the main ranking for need: it answers how well a council does, full stop, and these councils are where the two questions give clearly different answers.

Need is only partly captured by these measures, and adjusting is not an excuse: it says how a council does compared with others in similar circumstances, not whether its results are good enough for residents.

How reliable are the rankings?

We re-ran the rankings under reasonable alternatives: different pillar weights (50/25/25, 70/15/15), equal weight for every service, no bailout rule, leaving out temporary accommodation, and no cap on outliers. Rankings stayed closely correlated with the published version (rank correlation 0.98 or higher), the median council moved 4 places or fewer, and some councils moved up to 23 places. Giving the three pillars equal weight changes more (rank correlation 0.90, median move 11, up to 57 places), because value for money measures something different from service quality. So the bands (top quarter, upper middle, lower middle, bottom quarter) are a more reliable guide than exact positions.

Likely range for each council

Each council page gives a likely range for its overall rank. We re-ran the ranking 500 times. Each time, every service-area and pillar weight was multiplied by a random factor between 0.5 and 1.5, to reflect that reasonable people would weight things differently. Measures built from case counts (Ombudsman decisions, EHC plans, homelessness cases) were also re-drawn to reflect chance variation in small numbers. The range runs from the 5th to the 95th percentile of the council's rank across those re-runs, and always includes its published rank. We also show how often it stayed in the same band. Councils near the top and bottom are clear-cut; the median range spans 33 places of 130. The ranges don't cover errors in the source data or measures we don't have.

Who runs the council: shown, not scored

Each council page shows which party or parties run the council and how many seats each party holds. We don't score it: political control is not a measure of how well a council performs. The figures come from Open Council Data UK, a free dataset of council composition that its publisher releases as public domain; we credit it here. We check for a new release every month and the page says which file date the figures are as at, so a by-election or a change of administration shows up at the next monthly refresh, not instantly.

The publisher gives each council a short label. We show it in plain English and check it against the seat counts. A bare party (for example "Labour") means that party holds more than half the seats; if the seats don't support that, the refresh stops instead of publishing it. "No overall control" is followed by the parties in the administration, the leading party first and the others in order of size, which is the publisher's own rule. The publisher doesn't define two suffixes: we read "min" as a minority administration and "Mayor" as a directly elected mayor from that party. Those two readings are ours. We spot-checked a sample of councils against their own websites, not all 130. If a council's entry is out of date or wrong, please use the corrections page.

Audit status: shown, not scored

Each council page shows whether the council's accounts have a signed audit opinion, and when it was signed, from the monthly data published by Public Sector Audit Appointments (PSAA), using PSAA's own status wording. We don't score it. In 2025 and 2026 the government set "backstop" deadlines to clear a national backlog of unfinished audits, and most of the councils we rank carry PSAA's "delivered by backstop" status. Their opinions were signed on very different dates, from months before a deadline to the deadline itself, and the published data records when an opinion was signed, not which type it was. So we don't say whether any opinion was unmodified, qualified or disclaimed. Timing reflects auditor capacity as much as anything the council did, so scoring it would tell residents little. We also show how many of the last three audit years have no signed opinion at all; that does mark a council out, and is worth asking about.

What we don't score, and why

Some things residents care about are missing from the ranking. We checked each one against the official data and left it out because the figures can't compare councils fairly. A gap is better than a number that misleads.

• Roads. Each of these councils looks after most of the roads in its area, but the Department for Transport's road condition figures can't yet compare them fairly. Residential streets, where most potholes are, have no standard survey method, and half the London boroughs have no usable figures for B and C roads. For A roads, 28 of the 130 used survey methods that the Department says aren't directly comparable with the rest, and most councils have between 2% and 7% of their A roads in poor condition, a range in which the Department warns that results vary between survey machines. Road condition also builds up over decades of funding, much of it government grant, so it says little about how the current council runs the service. The figures are on GOV.UK.
• Crime. Policing is run by the police and crime commissioner, or in London, Greater Manchester, West Yorkshire, South Yorkshire and York and North Yorkshire by the mayor, not by the council. Recorded crime also follows how many people pass through an area, so city centres look worst whatever the council does. Local figures are on police.uk.
• Fly-tipping. Councils clear it and take action against it, but the official figures mostly show how each council counts it: 15 of the 130 record only incidents reported by the public, and busy central boroughs record far more. Enforcement figures mostly reflect each council's policy on fines. DEFRA, which publishes them, says councils shouldn't be classed as "good" or "poor" on these numbers. The figures are on GOV.UK.

We'll look again when better data is published, and any change will be in the changelog.

Corrections and challenges

If you think something is wrong, tell us. How we handle corrections, challenges and statements from councils is on the corrections page, and every change we make, with its date, is in the changelog.

Reproducibility

Every published number is traced to the exact source file it came from. This page describes methodology v2.0 applied to data snapshot adc5569776f7.