The university that discovered accessibility three weeks before enrolment

The situation

The European Accessibility Act applied from 28 June 2025, and it reaches UK organisations that offer services to customers in the EU. The presumed standard is EN 301 549, and version 4.1.1, incorporating WCAG 2.2, was published on 2 September 2026. New services launched after the June 2025 date had to comply from the outset; pre-existing services carry a five-year transition. Penalties run to several million euros.

A UK university recruiting across the EU is squarely in scope. Its application portal is a service offered to EU citizens, and the students least well served by an inaccessible portal are precisely the ones whose access it is supposed to widen.

The problem

The university had commissioned an accessibility audit, which had returned around six hundred pages of findings across an estate of roughly 2,400 pages. It was thorough, accurate and unusable.

Findings were listed by severity, which sounds sensible and is not. A critical finding on a 2011 conference microsite and a critical finding on the application form appear identical in a severity-sorted list. The team had been working through it in order and had spent six weeks on pages almost nobody visits.

Enrolment was three weeks away. The portal was in scope, definitely non-conformant, and nobody could say how badly, because the audit had covered it at the same depth as everything else.

What we did

We proposed ignoring most of the report. That is an awkward thing to recommend about a piece of work the client has just paid for, and it was the right call: the report was an inventory, not a plan, and treating an inventory as a plan is how six weeks went into a conference microsite.

We reframed the question from pages to journeys. A student does not visit a page, they complete a task: find a course, apply, upload documents, accept an offer, enrol, pay. We identified twelve journeys that carry essentially all of the consequential traffic and tested those end to end, with assistive technology, rather than against a checklist.

Testing with a screen reader rather than a scanner changed the findings substantially. The application form passed most automated checks and was close to unusable in practice: the multi-step progress indicator was invisible to a screen reader, so a user could not tell which of six steps they were on, and the document upload gave its only error feedback through colour.

Neither was flagged as critical by the audit. Both would have stopped a student applying.

We fixed the twelve journeys first, with the application and enrolment paths ahead of enrolment week, then handed over a prioritisation method rather than a list, so the remaining estate could be worked through by consequence instead of by severity.

The outcome

The application and enrolment journeys reached WCAG 2.2 AA conformance five weeks after we started, ahead of enrolment.

Twelve journeys were remediated and verified with assistive technology, not only with automated tooling. The wider estate is being worked through on the same basis, journey first, with the conference microsites last.

The university's own framing afterwards was the useful one. It had been treating accessibility as a document to produce for a regulator. What changed was treating it as whether a partially sighted applicant can actually get an offer accepted, which is both the legal test and the thing the institution says it exists to do.

An audit tells you every window that does not close. It does not tell you which one is above the server room.