Accessibility
What we have actually measured.
This page says what was tested, when, what it found, and what it does not cover. The numbers below are read from the last recorded test run rather than typed here, so they cannot say one thing while the site does another.
- Last measured
- 13 September 2026
- Tool
- axe-core 4.13.0
- Pages tested
- 10
- Renderings
- 30
- Widths
- 1440px, 390px
- Serious or worse
- 0
- Minor
- 0
Each page is tested at both widths and again with reduced motion, because the phone sheet, the desktop bar and the settled state are three different trees. Every page outside the Journal is opened; the Journal is sampled, because its articles all render through one layout. The exact pages are listed below rather than summarised.
//accessibility//contact//how-it-works//privacy//terms//journal//journal/catch-a-copycat-domain-before-it-goes-live//journal/category/playbooks//404.html
What the site commits to
- Every control that moves for more than five seconds has a visible pause, and remembers the choice.
- Nothing animates when the system asks for reduced motion, and every still frame is a composed one.
- Text meets WCAG AA against its own background, checked as a pair rather than assumed.
- Every interactive element is reachable and operable by keyboard, with a visible focus ring.
- The pages carry real landmarks and one h1, so a screen reader can skim them.
Where it still falls short
This list is written by hand, because no tool reports any of it. It is worth more than the score above.
-
The carousel is a complex widget
The row of copycat shapes on the homepage is a perspective carousel. It takes arrow keys when focused, it pauses on hover and on a real button, and the cards behind the centre one are hidden from assistive technology rather than read out. It is still the most complex thing on the site, and a simple list would be easier to operate.
-
The film is described, not captioned
The hero plays a silent film, so there is no speech to caption. It carries a text description of the whole sequence instead, and it does not start at all when the system asks for reduced motion: the still frame is the finished composition, not a blank box.
-
Automated testing finds roughly a third of barriers
The run above is axe-core. It is good at contrast, names, roles and structure, and it cannot tell whether a heading order makes sense, whether an error message is useful, or whether a keyboard path is sane. Those are judged by hand here, which means they are judged by the people who built it.
-
It has not been tested with a screen reader by someone who uses one
The accessibility tree is checked, the landmarks and names are checked, and the pages have been driven by keyboard. That is not the same as being used by somebody who relies on a screen reader daily, and it should not be presented as if it were.
If something blocks you
Tell us what you were trying to do and what stopped you, and we will fix it. A report that something is unusable is more useful than any audit, and it goes to the same place a review request does.