The Gloucestershire Website Accessibility Report

For many people, especially accessibility professionals, we have been using the same data for a very long time to report back to customers and the general public on accessibility issues. Reports that base accessibility against the top million websites online are hugely valuable data we can use, but they don’t always feel relatable. We have decided to break this down into smaller, more understandable reports, starting with Gloucestershire. We plan to analyse the websites that are most commonly used in each community: public sector websites, the major charities in the region and organisations within the county. These included some of the largest in the county, but we also wanted to test some organisations whose website accessibility would be perceived as having a more direct impact, for example serving customers with low vision.

The assessment was carried out using our own Squee Website Accessibility Checker. Each of the 50 websites was marked against the same twelve-point test, drawn from a subset of the WCAG 2.2 AA guidelines and generated through a combination of automated scanning and a small amount of manual testing, including navigating each site by keyboard.

The twelve checks cover the most common and most easily tested accessibility issues rather than every requirement in the guidelines, so this report is in no way a definitive result on the accessibility of each site. A conclusive audit would require a minimum of three to five hours on every site, which would not be feasible across a sample of this size. Marking every site against the same twelve checks is the most effective way of giving representative data.

Each website receives a single score out of 100, combining the automated assessment with the manual keyboard testing. The score is not a pass or fail mark. It is an indication of how much stands between a disabled visitor and the content, and it falls into one of five bands.

Each website was marked against the same twelve checks. Together they cover the barriers we see most often, and each can be tested reliably across a large sample.

1. Links without clear text. Every link should make sense on its own. Many screen reader users browse a page as a list of its links, so a link that reads “read more”, or nothing at all, tells them nothing.
2. Keyboard operation. Everything a mouse can do should also be possible by keyboard, including menus, forms and search.
3. Colour contrast. Text has to stand out from its background clearly enough to be read by someone with low vision, measured against a minimum contrast ratio.
4. Page landmarks. Pages should be divided into labelled regions, such as navigation, main content and footer, so assistive technology can jump straight to each one.
5. ARIA labelling. Buttons and controls need an accessible name. Without one, assistive technology can only announce “button”, leaving the visitor to guess what it does.
6. Heading order. Headings should describe the structure of the page in a logical order, because many assistive technology users navigate by jumping from heading to heading.
7. Visible focus indicator. As a keyboard user moves through a page, the selected element should be clearly highlighted so they can see where they are.
8. Form field labels. Every form field needs a label properly associated with it, so it is clear what should be entered and assistive technology can read it out.
9. Image descriptions. Images that carry meaning need a written alternative for people who cannot see them.
10. Page language. The language of a page should be declared in its code, so screen readers pronounce the content correctly.
11. Text size. Text should be large enough to read and should scale cleanly when a visitor enlarges it.
12. Media alternatives. Video and audio need captions or transcripts so the content is available to people who cannot hear or see them.

The average score was 52 out of 100, which falls in the “Below Average Accessibility” band of the assessment’s five-band scale, and 40 of the 50 sites fell into the two lowest bands. The clearest finding was that the size or wealth of an organisation was a poor guide to how accessible its website was. The charities averaged 50 out of 100, and the private organisations 48, almost identical despite very different resources. The only group to perform consistently well was the public sector, comprising councils, an NHS trust and universities, which is legally required to meet defined accessibility standards. Public bodies averaged 74.

Across all 50 sites, only five worked without problems when operated by keyboard in the elements we tested by hand, approximately 64 per cent gave no visible indication of the selected element while navigating, and we recorded more than 850 individual instances of text with insufficient colour contrast.

On this scale, the study average of 52 sits in the “Below Average Accessibility” band. More telling is the distribution: only six of the 50 sites reached “Good Accessibility” or better, while 40 fell into the two lowest bands, and 26 were rated “Poor Accessibility”.

It might be assumed that a large, well-funded organisation would maintain a more accessible website than a small charity operating on limited resources. Across these 50 sites, that assumption did not hold. Grouping the sites into three types shows the point clearly: the single reliable predictor of a good score was whether an organisation was under a legal obligation to meet accessibility standards.

Public sector: average 74 (6 sites)

Councils, an NHS trust and universities: the highest scoring group, and the only one to perform consistently. Public bodies have specific accessibility regulations to meet and are audited against them, which is the most likely explanation. Even so, this finding surprised us. These websites carry a legal obligation to meet the WCAG 2.2 AA standard, yet every one of the six still showed accessibility issues in our checks. Half showed problems when operated by keyboard, and two lacked a visible focus indicator, issues sites under a legal duty should not have at all. Scores still ranged from 41 to 95.

Charities and community groups: average 50 (24 sites)

The widest range of any group, from 22 to 90, spanning the highest charity score and some of the lowest scores in the study. None of the 24 charities we tested by hand worked without problems when operated by keyboard, and around seven in ten gave no visible focus indicator. Fourteen of the 24 fell into the lowest band. These are frequently the organisations whose visitors are most likely to depend on accessible design, which makes the spread the most consequential finding in the study.

Private organisations: average 48 (20 sites)

The county’s largest companies, together with independent institutions, clubs and attractions. Despite substantial budgets, this group scored marginally below the smaller charities, and eleven of the 20 fell into the lowest band. Several of the county’s most recognisable names sat at the very bottom of the sample, while a few scored well, so within the group the determining factor was evidently attention rather than resources.

1. Legal obligation predicted accessibility more reliably than budget

The public bodies were the highest scoring group, at 74, and the only group subject to specific accessibility regulations. Both non-public groups scored well below them and almost identically to one another. No public sector site blocked a keyboard user at any point, while 15 sites elsewhere did. Accessibility appears to need an accountable owner: in the public sector the law provides one, but elsewhere it is discretionary.

2. A high score did not mean the site worked by keyboard

Of the eight sites scoring 75 or higher, five still showed problems when operated by keyboard. At the other end of the scale, one of the five sites with faultless keyboard operation recorded one of the lowest scores in the study. A site can pass most of the automated checks and still frustrate a keyboard user, or work faultlessly by keyboard while failing them, so neither kind of testing is sufficient on its own.

3. Insufficient colour contrast was close to universal

Insufficient colour contrast appeared in every group, and almost every site we could measure failed on it: more than 850 individual instances across the sample. It is also typically among the least expensive issues to remedy. If there is one action every organisation in this study could take immediately, it is to check that text and controls meet the minimum contrast ratio.

4. An organisation’s purpose told us little about its website

The charities established specifically to support disabled people showed the widest range of scores of any group, from 26 to 90. Within every group, organisations of similar standing landed at opposite ends of the scale. The consistent differentiator was not what an organisation existed to do, but whether its website had ever been tested without a mouse, and the results suggest most had not.

Keyboard operation and colour contrast were the most severe and most widespread issues, but problems appeared across all twelve checks. The chart below shows the share of sites on which each issue was identified.

Most sites failed several of the twelve checks simultaneously: accessibility is not a single item to address, but a set of distinct requirements, each of which needs to be reviewed.

1. Links without clear text. The most widespread issue: nearly every site contained empty links, or links that convey no information in isolation.
2. Keyboard operation. Only five of the 50 hand-tested sites worked without problems; on 15, a keyboard user could not reach or complete parts of the page.
3. Colour contrast. Almost every measurable site contained text below the minimum contrast ratio, more than 850 instances in total.
4. Page landmarks. Missing on a large majority of sites.
5. ARIA labelling. Frequently absent or misapplied, leaving controls without an accessible name.
6. Heading order. Around seven in ten sites lacked a clear top-level heading or used heading levels inconsistently.
7. Visible focus indicator. Missing on 64 per cent of sites.
8. Form field labels. Unlabelled or incorrectly associated fields on roughly half the sites with forms.
9. Image descriptions. Missing or incomplete on a substantial share of sites.
10. Page language. Not declared on a minority of sites.
11. Text size. A minority of sites used text that was too small or did not resize cleanly when enlarged.
12. Media alternatives. Few sites used video or audio; where present, captions and transcripts were generally absent.

A significant number of people navigate the web without a mouse, including people with motor conditions, people with a tremor, and people with limited vision who use assistive technology. Each site was tested by hand, navigating by keyboard without a mouse. Of the 50 sites assessed this way, only five worked without problems. A further 30 could be operated, but with problems such as focus moving through the page in an illogical order. On 15, a keyboard user could not reach or complete part of the page. On 32, the selected element was not visibly indicated. Of the 45 sites with a drop-down menu, 16 could not be opened using the keyboard at all.

Insufficient colour contrast, such as pale grey text on a white background or brand colours placed on similar brand colours, is one of the most common reasons text cannot be read, and one of the least expensive to correct. Of the sites where contrast could be measured reliably, almost every one failed. Across the 50 sites we recorded more than 850 individual low-contrast areas, with the highest single count exceeding one hundred.

These are not careless organisations or poorly built sites. Many are well designed, well resourced and clearly maintained. Contrast and keyboard operation are simply the aspects that do not become apparent when a site is reviewed with a mouse and full vision, so they are missed, on charity sites and prominent organisations alike, unless they are specifically tested for.

Approximately one in five people in the UK is disabled. When a website is difficult to use for them, the consequence is not a lost sale but a person who needed something and did not obtain it: a supporter who abandons a low-contrast donation form, a patient who cannot reach the relevant page by keyboard, a customer who cannot read the price. For a charity, the barrier often affects precisely the people it exists to support. For a business, it is customers and revenue that are lost without ever being recorded.

There is also a legal dimension. The Equality Act 2010 requires service providers to make reasonable adjustments, and that duty applies to websites as it does to physical premises. Public bodies are additionally subject to the Public Sector Bodies Accessibility Regulations 2018, the most likely reason they were the strongest-performing group in this study.

The homepage of each of the 50 organisations was assessed in July 2026. Every page was evaluated using an automated scan built on the WAVE and axe accessibility engines, which test colour contrast, image descriptions, headings and page structure against the WCAG 2.2 AA guidelines. Each site was additionally tested by hand, navigating by keyboard without a mouse.

Automated tools identify approximately half of all accessibility issues, so the automated figures here should be regarded as a minimum. Where an automated contrast check could not be completed reliably, that site is excluded from the contrast figures rather than recorded as passing. The 50 organisations comprised 24 Gloucestershire charities and community groups and 26 of the county’s largest organisations, grouped into three categories: public sector bodies, charities and community groups, and private organisations. The sample was selected rather than drawn at random, so figures describe the 50 sites assessed rather than a statistical estimate for the county as a whole. Findings are reported in aggregate and anonymised form, and no individual organisation is identified.

This report describes patterns across a sample. It is not an assessment of any single organisation, and none is named. Organisations wishing to know how their own website performed are welcome to request a private, complimentary summary.