Accessibility of this site

WCAG 2.2 Level AA across every public page, and AAA for text contrast wherever the design allows it. Colour pairs on this site are measured rather than picked by eye, and most sit well above the AA threshold.

We are not claiming conformance. Claiming it is exactly the move we tell people to distrust, and clause 03 of the terms says we will never issue a certificate — including to ourselves.

• Every colour was measured, not chosen. The palette was run through this product’s own contrast engine, the same code that measures your site.
• Severity is never carried by colour alone. Every band is a word, a shape and an ink, so nothing on this site depends on telling red from amber.
• Interactive things are the real elements. The findings browser is a tablist with arrow-key navigation on a wide screen and an accordion of headings and buttons on a narrow one — never a set of clickable divs, which is precisely the defect this product charges to find.
• Reduced motion removes animation rather than shortening it. Somebody who asked their system to stop moving things wants it stopped, not hurried.
• Scroll regions announce themselves. Anything that scrolls is keyboard-reachable, has a name, and draws a visible bar at the same 3:1 a form field owes.

A statement with nothing in this section means nobody looked.

• A review of our own dashboard in August found real defects, now fixed. Controls that removed themselves when pressed — “Show the other places”, “Undo”, Download — dropped keyboard focus to the top of the page. Every control that could not be pressed was dimmed to about 1.6:1, which made the label explaining why the least readable thing on the screen. Filtering and paging the work list replaced every row with nothing announced. The trend chart’s data table sat in a collapsed disclosure a keyboard could not scroll, and its axis labels rendered at about five pixels on a phone.
• More is open than is closed. That review produced a list longer than the part fixed so far, most of it about keyboard and focus behaviour inside the signed-in dashboard. The public pages and the scan report have had the most attention; the dashboard has had the least.
• Manual testing is ongoing. Automated checking reaches part of the problem and we say so on the home page rather than pretending otherwise; testing with assistive technology across every page and browser continues alongside it.
• Testing with disabled users is planned and is the work we most want to do next.
• The specimen on the home page is a drawing of a bad page. Nothing in it is a working control — the menu, the link and the sign-up field are spans, deliberately, so we never ship an unlabelled field in order to illustrate one. It is announced as a figure with a caption saying so.
• PDF reports are not produced, and PDF checking is not implemented, so if a PDF is how you would rather read this we do not have one. The downloadable report is a single HTML file.

Tell us. Email [email protected] and say what you were trying to do, what happened, and what you were using. You do not need to know the WCAG number, and you do not need to be polite about it.

We aim to reply promptly and to say plainly whether we can fix it, when, or why not. A barrier on this site is treated as a bug of the highest severity we have.

It is dated, and it names real gaps rather than only the parts that pass.