VDS is built to WCAG 2.1 AA, in line with the Public Sector Bodies (Websites and Mobile Applications) Accessibility Regulations 2018. That’s the legal baseline for anything Vale publishes — but the more useful way to think about it is simpler: as many people as possible should be able to use what we build, regardless of how they’re using it.
In practice, that means someone should be able to change colours, contrast, and font size without anything breaking; zoom in up to 300% without content spilling off the screen or becoming unusable; navigate entirely by keyboard or by voice; and use a screen reader, including current versions of JAWS, NVDA, and VoiceOver, to understand and act on every part of a page.
It also means writing in plain language wherever possible — Vale’s work involves a lot of scientific and Latin terminology, and where that’s unavoidable, content should explain it in plain English rather than assuming familiarity.
Why this sits at the component level
Most accessibility failures aren’t one-off content mistakes — they’re the same structural problem, repeated everywhere a pattern gets reused: a custom dropdown with no keyboard support, a form error that only shows as a colour change, a focus ring that’s been styled away. When that pattern lives in a shared component library, the fix (or the failure) happens once and applies everywhere the component is used, instead of being solved, or missed, separately by every team that builds a page.
That’s the actual job of VDS here: get these things right once, at the component level, so product teams inherit accessible behaviour by default rather than having to rebuild it — or accidentally break it — on every project.
What every component has to get right
These aren’t aspirational — they’re the specific, recurring ways components fail assistive technology users if they’re not handled deliberately:
- Visible focus. Every interactive element needs a clear, visible focus indicator when navigated to by keyboard. Removing the default outline without replacing it is the single most common way a component becomes unusable for keyboard users.
- Contrast. Text meets 4.5:1 against its background at minimum; interactive elements, borders, and non-text UI (focus rings, input outlines, icon-only controls) meet 3:1. This applies to every state, not just the default — disabled, hover, and focus states have historically been the ones that get missed.
- Real semantics, not just visual mimicry. A custom accordion, tab, toggle, tooltip, or date picker has to actually behave like one to assistive technology — correct roles, states, and keyboard interaction — not just look like one visually. A
<div>styled to resemble a button isn’t a button until it’s operable and announced as one. - State is programmatic, not just visual. Disabled, expanded, selected, and current states need to be exposed to assistive technology, not only conveyed through colour or opacity. If a screen reader user can’t tell a button is disabled, colour alone didn’t communicate it to them.
- Descriptive link and button text. “Click here” and “read more” fail on their own — link and button text should make sense out of context, since that’s often exactly how a screen reader user encounters it, in a list of every link on the page.
- Errors are identified, not just shown. A validation error needs to be associated with its field, announced to assistive technology, and stay visible until it’s resolved — not disappear after a few seconds, and not rely on colour alone to indicate something’s wrong.
- Logical, complete keyboard access. Every interaction available with a mouse needs a keyboard equivalent, in a focus order that matches the visual layout, with nothing that traps focus inside a component.
- Language and input hints are marked up. Foreign-language terms (Latin plant names are the obvious case at Vale) get a
langattribute so screen readers pronounce them correctly rather than reading them as English. Fields collecting personal information carry the rightautocompletevalues so browsers and assistive tools can fill them in reliably.
None of this is unusual or Vale-specific — it’s the standard set of failure modes WCAG 2.1 AA exists to catch.
The point of listing them here is that they’re exactly the kind of thing that’s cheap to get right once in a shared component, and expensive to retrofit later across every product that already shipped with the broken version.