We scanned 54 UK council websites. Most are excellent. That's the story.
July 2026 · AllyRadar research · 8 July 2026
Public-sector websites have been legally required to meet WCAG 2.1 AA since September 2020, with mandatory accessibility statements and central monitoring. Last week we scanned the UK's biggest retailers — where no such regime yet applies — and 75% failed. So we ran the same scan across 54 of the UK's largest local councils. The contrast is the finding:
Regulation works
Same automated checks, same engine, same month. Commercial sites with no enforcement: 75% fail. Public-sector sites with a legal mandate and monitoring: 26% fail — and Wales recorded a clean sweep, every council scanned scoring 100. This is the strongest real-world evidence we've collected that accessibility isn't hard — it's simply done where someone is watching, and not done where nobody is. The European Accessibility Act is now bringing the watching to e-commerce.
The councils still failing the legal standard
The regulations have applied to these sites since 23 September 2020. Six years on, 14 of 54 scanned councils have at least one serious or critical WCAG failure on their homepage. (A note on what that does and doesn't mean: the regulations allow councils to declare “disproportionate burden” exemptions in their accessibility statements, so a WCAG failure is not automatically unlawful — but the failures below are basic template issues of the kind such exemptions are rarely, if ever, claimed for.)
| Council | Nation | Score | Critical | Serious | Most impactful issue |
|---|---|---|---|---|---|
| Armagh Banbridge Craigavon | NI | 22 | 1 | 4 | Links must have discernible text |
| Wolverhampton City Council | England | 39 | 1 | 1 | ARIA attributes must conform to valid names |
| Peterborough City Council | England | 51 | 1 | — | Buttons must have discernible text |
| Plymouth City Council | England | 51 | 1 | 2 | Images must have alternative text |
| Barnet Council | England | 72 | — | 4 | Elements must meet minimum color contrast ratio thresholds |
| Fermanagh and Omagh | NI | 72 | 1 | — | Buttons must have discernible text |
| Ealing Council | England | 78 | — | 3 | Elements must only use permitted ARIA attributes |
| Southampton City Council | England | 78 | — | 3 | Documents must have <title> element to aid in navigation |
| Doncaster Council | England | 85 | — | 1 | Links must have discernible text |
| Hounslow Council | England | 85 | — | 2 | Documents must have <title> element to aid in navigation |
| Birmingham City Council | England | 92 | — | 1 | <html> element must have a lang attribute |
| Medway Council | England | 92 | — | 1 | <html> element must have a lang attribute |
| Stirling Council | Scotland | 92 | — | 1 | <html> element must have a lang attribute |
| Tower Hamlets Council | England | 92 | — | 1 | Links must have discernible text |
The most common failures mirror the retail study almost exactly: missing lang attributes, insufficient colour contrast, and links with no discernible text — one-line fixes, unmonitored. Council websites are how residents apply for housing, report repairs, and claim support; the 40 councils scoring 100 prove there's no technical excuse.
Download the full dataset — all 54 councils (CSV) →
Methodology — and honest limits
- Homepage-only scans of 54 of the UK's largest local authorities (all four nations), 8 July 2026, using a real Chromium browser and axe-core against WCAG 2.1 A/AA. Scores are AllyRadar's severity-weighted 0–100 scale; the scoring formula is open source.
- Automated testing catches a large share of WCAG failures but cannot verify every criterion — a perfect automated score is necessary, not sufficient, for compliance with the Public Sector Bodies Accessibility Regulations 2018. Failures, however, are failures by definition.
- Homepage-only is a light-touch sample: deeper journeys (forms, payments, planning portals) are where council sites typically struggle most.
- Fixed your issues? We'll happily re-scan and update this page — email research@allyradar.app.
Is your site watched?
The lesson of this study: monitored sites stay accessible. Scan yours free — score, failing code, and an AI fix prompt for every issue, in about 20 seconds.
Scan your site free →This study may be reproduced with attribution and a link to allyradar.app.