Skip to content
UXE Award

Accessibility scoring: exactly what editors check, in the order they check it

Accessibility is one of the seven criteria on every UXE entry, in every category. The tests editors run in the first ten minutes, what an honest conformance claim looks like, and why disclosed gaps raise a score.

The UXE editors6 min read

Accessibility is one of the seven criteria, scored on every entry, in every category — not a specialist award you can opt into. A product can be beautiful, fast and well-structured and still lose a point and a half here, which is usually the difference between Featured and Official Selection.

This is what editors actually do, in roughly the order they do it. None of it takes specialist equipment; all of it takes about ten minutes.

The first ten minutes

Four tests, ten minutes
  1. 01

    Unplug the mouse

    Tab through the main task. Is focus visible, ordered, and never trapped?

  2. 02

    Zoom to 200%

    Does the layout reflow, or does content disappear off the side?

  3. 03

    Turn on the screen reader

    Are controls named? Are images described? Does anything announce itself as "button"?

  4. 04

    Check the colours

    Contrast on body text, on placeholder text, on the disabled state that still has to be read.

Keyboard: where most entries lose their first point

  • A visible focus indicator on everything. Removing the outline without replacing it is the single most common accessibility failure in award entries.
  • Focus never obscured. A sticky header that covers the focused field is a WCAG 2.2 failure and an everyday annoyance.
  • Focus order that follows the visual order. If the tab sequence jumps around, the DOM and the layout have diverged.
  • No traps. Dialogs, embedded players and custom widgets must be escapable — and must return focus where it came from.
  • Skip link. Still the cheapest accessibility feature in existence, still missing from most entries.

Screen readers: names, roles, and the announcements nobody tested

You don't need to be an expert user to find the big problems. Turn on VoiceOver or NVDA, tab through the main flow, and listen for three things: controls with no name ("button"), state that is never announced (a toggle that doesn't say on or off), and dynamic changes that pass silently (a filter that updates results with nothing said).

Six announcements
Compared onAnnounced wellCosts points
Icon button"Delete invoice 4417, button""Button"
Toggle"Email notifications, switch, on""Checkbox"
Results update"24 results" announced politelySilent change
ErrorAnnounced, and focus moves to the fieldRed text nobody hears
Loading"Loading results", then the countSilence, then content
ModalTitle announced, focus moved in, background inertFocus still behind the dialog

Colour, contrast and the things that fail quietly

WCAG 2.2 AA asks for 4.5:1 on body text, 3:1 on large text and on the visual boundaries of user-interface components. In practice, the failures cluster in the same five places every time:

  • Placeholder text used as a label — usually far below 4.5:1, and it disappears when typing starts.
  • Disabled states that still need to be read. Disabled does not mean invisible.
  • Text over photography, where contrast depends on which image loaded.
  • Status indicated by colour alone — red/green, with no icon, no label, no pattern.
  • Focus indicators that meet contrast against one background but not the other.

Motion, timing and targets

Three WCAG 2.2 additions come up constantly in modern entries. Reduced motion, honoured — if the operating system asks for less motion and a parallax hero ignores it, that's a failure with a physical cost for some people. Target size, at least 24×24 CSS pixels for interactive controls. And dragging: anything you can drag must also be achievable without dragging.

The conformance claim: honesty beats ambition

Entries declare a conformance level. Editors treat that declaration as a claim to be tested, and the scoring consequences are asymmetric — which surprises people.

How claims are scored
  • Best

    Partial claim, gaps disclosed, fix dates given

    Trust rises across every other claim in the entry. This scores highest, even against a stronger claim.

  • Good

    Full AA claim, audited, holds up under testing

    Exactly what it says. Verified in the first ten minutes.

  • Weak

    No claim, nothing tested

    Scored on what we find, which is usually less than the team hoped.

  • Worst

    Full claim that fails a keyboard check

    Costs more than claiming nothing. It puts every other claim in the entry in doubt.

A disclosed gap is a team that has looked. An undisclosed one is a team that hasn't — and we will find it in ten minutes, which tells us which it was.
From the accessibility criterion brief

What to send

  • Your accessibility statement, with real findings rather than a template.
  • An audit if you have one, including the issues that are still open.
  • A note on what you tested with: which screen readers, which browsers, which real users.
  • The parts you know are weak, and the plan. This costs you nothing and earns you trust.

Inclusive & Ethical Design requires a statement; every other category scores accessibility anyway. If you want one independent read on where you stand, entering is free — and the scorecard names the criterion. Submit a project.

Automated tools: useful, and not enough

Automated checkers are worth running — they catch missing labels, contrast failures and structural problems cheaply. What they cannot do is tell you whether the reading order makes sense, whether a custom widget behaves the way a screen-reader user expects, or whether an error message is understandable.

A clean automated report is the floor, not the ceiling, and an entry that cites one as proof of accessibility tells editors that nobody used the product with assistive technology.

Where custom components go wrong

Six widgets, six ways they exclude
Compared onCommon failureWhat it costs the user
Custom selectNot operable by keyboard, state not announcedCannot choose an option at all
ModalFocus stays behind it, background not inertTabs into a page they cannot see
ToastAppears and disappears silentlyNever learns the action worked
Infinite scrollNo landmark, no way to reach the footerTrapped in a list forever
Drag-and-dropNo alternativeCannot perform the task
CarouselAuto-advances, cannot be pausedLoses their place repeatedly

Cognitive accessibility, which almost nobody enters

Most accessibility work targets vision, hearing and motor ability. The largest under-served group is cognitive: memory, attention, literacy, anxiety. The design moves are unglamorous and effective — plain language, one task per screen, no unnecessary time limits, consistency across the product, and never relying on the user remembering something from three screens ago.

An entry that addresses this directly stands out, because it is rare and because it helps everyone.

Questions we get asked

  • Is accessibility scored in every category?

    Yes — it is one of the seven criteria on every entry. You cannot opt out of it, and a strong score elsewhere will not compensate for a weak one here.

  • What conformance level should we claim?

    The one you actually meet, with the gaps named. A partial claim with disclosed gaps scores better than a full claim we can disprove in ten minutes.

  • Do we need a third-party audit?

    No, but it's the strongest evidence available and unusually persuasive when you include the open findings as well as the fixed ones.

  • Does dark mode or high contrast mode count?

    They help, if they're done properly. A dark mode with contrast failures costs points rather than earning them.

  • What's the fastest accessibility improvement before entering?

    Restore a visible focus indicator on every interactive element, and label every icon-only button. Both are an afternoon's work and both are checked in the first two minutes.

Assessed work that scores highest on accessibility

Chosen by score from the work the editors have assessed, not by hand.

  • accessibility
  • wcag
  • scoring

Written by

The UXE editors

The team that selects the work, uses it, scores the seven criteria and writes every verdict on this site.