An accessibility audit is the usual way an organisation finds out where it stands. A specialist tests the site against the Web Content Accessibility Guidelines (WCAG), a report arrives with findings graded by severity, the developers work through them, and a year later somebody commissions the next one. Measure, fix, measure again. On paper, that looks like a programme.
However, the second audit keeps finding the first audit’s problems.
The audit did its job both times. The failure was in everything that happened after it. Nobody owned the report once it landed, and the next year’s budget went on another inspection instead of the skills that would have stopped the same problems coming back.
Lasting change needs a proper handover of the findings, then a deliberate shift of budget from buying audits to training the people who make the product. That gap between the report and the habit is where our accessibility work starts.
The legitimate role of an accessibility audit
Audits earn their place. Each one gives an organisation an independent baseline, written in the vocabulary regulators and procurement teams already use. Guidance from the Government Digital Service tells teams whose services face a Service Standard assessment to get one before public beta, and to fix what it raises first.
For a business that has never tested with assistive technology, the ranked list it produces is often the first hard evidence anyone has seen. That makes an expert audit, with help fixing what it finds, a sensible place to start.
The limits are just as plain. Testers sample, because checking every screen of a large service is rarely practical. An audit is a snapshot of what was found on the day, while releases, content and components keep arriving after the testers leave.
Findings also describe symptoms. A missing label on a shared form component shows up on every page that uses it, so one defect gets written up many times over.
Audits measure. Sheri Byrne-Haber calls them lagging indicators, because they judge decisions after those decisions have been made. Asking one to change behaviour is where programmes go wrong.
Why do accessibility issues keep coming back after an audit?
The WebAIM Million, a February 2026 scan of the most popular million home pages, detected WCAG failures on 95.9% of them, averaging 56.1 errors a page. That figure rose 10.1% on the previous year, reversing what WebAIM calls a trend of gradual improvement in recent years.
Low contrast text, absent alternative text and unlabelled form fields sit among six error types that have led the list for seven years running. Automated checks of home pages undercount by design, and the scan covers home pages across the web rather than sites that have been audited, so it shows the wider pattern. They still describe an industry fixing the same things again and again, a tolerance of inaccessibility that WebAIM’s Christopher Phillips reads as a cultural problem more than a technical one.
Sheri Byrne-Haber describes a large company that remediated its 100 busiest pages, watched them become inaccessible again within 18 months, then launched a new website that carried none of the lessons over.
The structures meant to prevent this mostly exist already. Level Access, in its State of Digital Accessibility vendor survey of 1,622 digital professionals in July and August 2025, found 77% reporting a policy, an accountable owner and a dedicated budget. Only 41% strongly agreed that their organisation approaches accessibility proactively. The paperwork is in place, and what’s missing is a route from a finding to a changed habit.
Why audit findings stall after the report lands
A survey of 160 accessibility professionals by Parthasarathy and colleagues, presented as an extended abstract at the leading international conference on human-computer interaction in 2025, asked how often designers and engineers act on specialist feedback. Only 10% said always. Nearly three quarters (74.4%) reported no success metrics for accessibility, and 51.9% had no mandatory accessibility review.
Set those figures beside the usual deliverable, a long document from outside the team that lives outside the backlog and lists instances instead of causes. No one’s objectives mention it, and nothing checks whether it was acted on.
Time works against it too. Karl Groves estimates from his firm’s own sales and delivery figures plus a poll of practitioners that around seven months pass between deciding to buy an audit and finishing the fixes. Plenty of releases ship in seven months.
Handing over an accessibility audit so the fixes last
The handover is a short piece of work, and it decides whether the audit pays for itself. At best, I see teams working with us to bring audit findings into their workflows, or asking us to log them there directly. They jump on calls with our accessible web developers to learn how to fix them, and sometimes we end up training their external development agencies too.
At worst, I’ve seen inaction. Mostly that comes from teams not being able or willing to take in our findings in a way that works for the people who have to act on them. The steps below are our recommendation, built on the evidence above, and no published trial has tested them as a set.
- Move every finding into the team’s own tracker: we think consultants should log issues there directly, because a finding left in a document stays unowned.
- Group findings by where they came from: nearly every one traces back to a shared component, a content habit, a third-party product or a single design decision.
- Give each source an owner: component defects belong to the design system team, and supplier defects to whoever manages the contract.
- Turn every fixed pattern into a check: that might be an automated test in the build pipeline or a line in the design review. Without one, a fix lasts until the next release.
- Re-test beyond the original list: as Bogdan Cerovac points out, retesting only the flagged items misses whatever the releases in between broke.
| Where the finding comes from | Who owns it | Prevention |
|---|---|---|
| Shared component or template | Design system team | A fixed component with an automated test in the build pipeline |
| Content and documents | Content lead and authors | Authoring guidance and a publishing checklist |
| Third-party product or plugin | Whoever manages the supplier contract | Accessibility requirements written into procurement and renewal |
| Design decision on colour, layout or interaction | Design lead | Accessibility step added to design review |
| One-off build error | The product team that shipped it | Code review that includes an accessibility check |
Vispero’s vendor analysis of accessibility regression lists unclear ownership, ungoverned design systems and content outpacing review among the causes. That’s why the table gives every source an owner.
The UK government’s monitoring team reports, on the Accessibility in government blog, that one audited organisation is writing accessibility into the procurement for its website redesign. Another found that relying on an outside company to make its changes slowed the fixes down. Both fix a cause rather than a single defect.
None of this should fall to the client alone. Stopping at the report hands the hard part back to the team least equipped to do it. Consultancies should go further and help teams make the fixes, through remediation consulting alongside the developers and accessible web development where a shared component needs rebuilding. That’s how we work after an audit, because a team that fixes the first instance with a specialist beside it can usually handle the rest without one.
Turning a fix into a rule
A reminder to use accessible headings depends on someone’s memory, and a checklist step before publishing adds a review point. A content management system (CMS) template that only offers semantic headings, and flags a broken structure, stops the mistake being made at all. The remedy is a better template, training for the people in that role, and a check before anything publishes. Whatever the fix, write it down as a requirement, with the point in the workflow where the check happens, an owner, the evidence and a measure, as in the example below.
| Part of the requirement | Example |
|---|---|
| Requirement | Every new or changed feature passes the accessibility checks the team agreed. |
| When | As part of the normal workflow, before anything is released. |
| Owner | The feature team, with the product owner recording any exceptions. |
| Evidence | The release record holds the check results and any blockers still open. |
| Measure | A rising share of issues caught pre-release, and fewer defects that come back. |
Why buying more audits won’t build an accessibility programme
Once the handover works, the next audit becomes a far better instrument. Its job shifts from rediscovering the same causes to checking whether the system holds. That changes what the budget should buy, because re-auditing every year without changing how you work means paying repeatedly for a list you already own.
The Accessibility Maturity Model from the World Wide Web Consortium, a Group Note published in November 2025, describes its highest level as accessibility “consistently evaluated, and actions taken on assessment outcomes”. Evaluation happens at every stage of maturity. Acting on it is what marks the top.
We organise our accessibility work around three parts that hold each other up, which are compliance, capability and culture. Compliance can be bought in. Capability and culture can’t.
A team either knows how to build an accessible date picker or it doesn’t, and no report settles that for them. So keep a smaller, targeted audit to validate the work, and move the rest of the money into training.
Accessibility training that changes how teams work
Most organisations already train, which is the uncomfortable part. In the same survey of 160 professionals, 92.5% said their organisation ran accessibility training, and of those, 41.89% said it came to less than 3 hours a year. Only 45% of Level Access’s respondents rated their training highly effective. Training exists nearly everywhere, but training that changes behaviour is scarcer.
Format looks like it matters more than presence. Parthasarathy and Joshi, in an experience report on a 12-week course for 66 software professionals at one multinational, compared a community of practice, meeting weekly for hands-on sessions, with a self-paced video group. On a practical audit exercise, the community group improved by 46.88 points and the self-paced group by 20.45. It’s one company and one cohort, so treat it as a signal that points the same way as everything else here.
This is where the handover pays twice. Every audit report is the best syllabus an organisation will ever be handed, because every finding is a mistake its own people made in its own product.
- Train by role: designers take the findings on contrast, interaction and focus order, developers take semantics and components, and content authors take headings, links and documents.
- Use your own findings as the exercises: a team repairing its own broken component learns more than one watching a generic demonstration.
- Learn from real users: watch disabled people attempt your journeys with their own assistive technology, and make each barrier they hit a training case. Few people forget watching a customer get stuck.
- Keep it going: short regular sessions and named accessibility champions, both suggested in the government’s Service Manual after an audit, outlast one annual course.
- Measure it: track training completion by role, how many champions are active in each team, and how many findings in the next audit repeat one from the last.
The question to ask before your next accessibility audit
Audits show where you stand. Whether next year’s report looks any different is the organisation’s decision alone.
So before signing off the next one, ask yourself this. Of the findings in your last audit, how many had already appeared in the audit before it, and what did you spend on stopping them appearing a third time?
We help teams keep their fixes in place after an audit, through role-based training, design system reviews and maturity evaluation. See how we help teams stay accessible.
Frequently asked questions
What is an accessibility audit?
An accessibility audit is an expert evaluation of a website, app or digital service against the Web Content Accessibility Guidelines (WCAG), usually at the middle conformance level (AA). Specialists combine manual testing, assistive technology such as screen readers, and automated scanning, then report each failure with a severity rating and a recommended fix. Testers work from a sample of screens and journeys, so the report describes the product at one point in time. Turning those findings into lasting change depends on the handover that follows.
What is a manual accessibility audit?
A manual accessibility audit is one in which a specialist tests pages and journeys by hand, using a keyboard, screen readers, magnification and other assistive technology alongside automated tools. Manual testing matters because scanners cannot judge many WCAG requirements, such as whether alternative text describes an image usefully or whether focus moves in a sensible order. Automated results on their own miss a large share of the barriers disabled people meet.
How much does an accessibility audit cost?
The cost of an accessibility audit depends on its scope, meaning how many templates, components and journeys are tested, which platforms are covered, how much of the testing is manual with assistive technology, and whether a re-test is included. Organisations should also budget for the work after the report arrives, such as moving findings into the backlog, fixing shared components and training staff. That later work decides whether the audit spend produces lasting change.
How do organisations remediate accessibility issues on websites?
Organisations remediate accessibility issues most efficiently by moving every audit finding into the team’s own issue tracker, grouping findings by the component, content process, supplier or design decision that caused them, and fixing each source once. Each source needs a named owner. Every fixed pattern should become a check, such as an automated test or a design review step, so later releases do not reintroduce it. A re-test should cover the whole sample again, including screens that were never flagged.
How do companies measure accessibility remediation progress?
Companies can measure accessibility remediation progress by counting how many findings in each new audit repeat a finding from the previous one, alongside issues closed by severity and time taken to fix. A falling repeat rate shows that causes are being removed. People measures belong beside them, such as training completion by role and the number of active accessibility champions in each team, along with a check on whether new releases introduce regressions.
What accessibility training do designers and developers need?
Designers and developers need role-based accessibility training that runs over weeks and uses their own product’s defects as practice material. Designers benefit most from work on contrast, interaction patterns and focus order, and developers from semantic markup, accessible components and testing with assistive technology. A 2024 experience report on 66 software professionals found a 12-week community of practice with weekly hands-on sessions improved scores on a practical audit exercise by 46.88 points, against 20.45 for self-paced video learning.
Related reading
- What digital accessibility covers, and who it affects
- How to evaluate your website’s accessibility, from automated checks to user testing
Sources
- WebAIM (February 2026) The WebAIM Million: the 2026 report on the accessibility of the top 1,000,000 home pages. Automated scan of home pages only.
- Christopher Phillips, WebAIM (30th April 2026) Tolerating Inaccessibility.
- Sheri Byrne-Haber (14th November 2024) When organizations do not bake accessibility reviews into its entire management process for both code and content, remediated digital properties are all but guaranteed to backslide into inaccessibility. Anecdotal.
- Sheri Byrne-Haber (20th April 2026) You can’t audit your way into accessibility culture change.
- Level Access (7th November 2025) State of Digital Accessibility Report 2025–2026: exploring our findings. Vendor survey of 1,622 professionals, carried out by Vanson Bourne in July and August 2025, mostly at organisations with more than 1,000 employees in the US, UK and EU. See also the methodology page.
- P.D. Parthasarathy, Rachel F. Adler, Devorah Kletenik, Swaroop Joshi and Anshu M. Mittal (2025) Skill, Will, or Both? Understanding Digital Inaccessibility from Accessibility Professionals’ Viewpoint. CHI 2025 extended abstract, survey of 160 professionals, fieldwork July to December 2024. The 41.89% training-length figure is a share of the 148 respondents whose organisation runs training.
- Karl Groves (9th November 2023, updated 21st October 2025) How long until your website is accessible? Built from his firm’s own median sales and delivery times and an informal poll.
- Karl Groves (2014, updated 22nd February 2023) Accessibility Consulting is Broken.
- Karl Groves (2011, updated 22nd February 2023) Prioritizing Remediation of Web Accessibility Issues.
- Bogdan Cerovac (10th December 2023) Potential problems with accessibility audits and what to do about them.
- Vispero (2nd July 2026) Accessibility Regression: Why Accessibility Breaks Down. Vendor content.
- Government Digital Service (last updated 22nd August 2019) Getting an accessibility audit. GOV.UK Service Manual.
- Keeley Talbot, Andrew Mitchell and Elisabeth Ellis, Government Digital Service accessibility monitoring team (23rd August 2023) How organisations respond to accessibility monitoring. Anecdotal, three organisations.
- World Wide Web Consortium, Accessible Platform Architectures Working Group (4th November 2025) Accessibility Maturity Model. Group Note, endorsed by the working group, not by W3C as a whole.
- Craig Abbott, TetraLogical (14th September 2026) Sustainable accessibility: compliance, culture, and capability, and Craig Abbott (2021) Defining a strategy for accessibility.
- P.D. Parthasarathy and Swaroop Joshi (2024) Teaching Digital Accessibility to Industry Professionals using the Community of Practice Framework: An Experience Report. arXiv, 66 professionals at one multinational, course run December 2022 to February 2023.