What works, what we measured, and what we have not.
We aim at WCAG 2.1 level AA on the public site. This page says what that means in practice, how it was checked rather than assumed, and the parts we have not tested yet. If something here does not work for you, one email fixes it faster than anything else on this page.
Last updated
The standard we aim at
Web Content Accessibility Guidelines 2.1, level AA, for the public pages: the homepage, the FAQ, the contact page and these legal pages. We also hold the WCAG 2.2 additions that apply to a site this simple, which in practice means target sizes and focus appearance.
That is a target and not a certificate. Nobody has audited this site for us, we have not been through a third-party conformance review, and this page does not claim a VPAT or an accessibility certification, because those are things a firm buys and we have not bought one. What we have is a measurement we ran ourselves and are willing to show.
What was measured, and how
Not by reading the code, which is how a site passes a linter and fails a person. Every public page was loaded in a real browser and inspected as rendered.
| What | How it was checked | Result |
|---|---|---|
| Colour contrast | For every element with text on it, the browser's computed text colour against the background actually composited behind it, including tints and textures. Compared to 4.5:1 for normal text and 3:1 for large. | Nineteen failures found and fixed on 17 September, most of them the brand green used at 10 to 12 pixels. No text on any public page is below the threshold now. |
| Keyboard | Tabbed through each page from the top, recording every stop in order. | Every control is reachable, the order follows the page, nothing traps focus, and every stop shows a visible focus ring. |
| Screen reader naming | Read out of the browser's own accessibility tree rather than guessed from the markup. | Every link, button, field and checkbox has a name. No unnamed controls. |
| Images | Every image on every page. | Each one either describes itself or is marked decorative and skipped. No image is left for a screen reader to read out as a filename. |
| Zoom and reflow | Loaded at a 320 pixel viewport, equivalent to 400% zoom on a laptop. | Everything reflows into one column and no page scrolls sideways. The two wide tables, on the privacy page and this one, scroll inside their own frame rather than pushing the page: that is deliberate, because a retention table loses its meaning if the columns are stacked. Both frames are reachable and scrollable from the keyboard. |
| Target sizes | Every clickable thing at that width, measured by where a pointer actually lands rather than by the size of the text: the hit area, including the invisible margin around a small link. | Nothing interactive has an effective target under 24 by 24 pixels. Links inside a sentence are exempt and were skipped. Several nav links and the contact form's checkbox were under before 17 September. |
| Motion | The stylesheet honours the reduced-motion setting your system already has, checked as a rule in the stylesheet rather than as an intention. | Turn it on and animation and smooth scrolling stop. Nothing here flashes, autoplays, or moves on its own whether you turn it on or not. |
| Structure | Headings, landmarks and page language, per page. | One first-level heading per page, no skipped levels, a main landmark, a skip-to-content link as the first thing a keyboard reaches, and the page language declared. |
What we have not tested
Said plainly, because a statement with no gaps in it has not been checked.
- No real screen reader, driven by somebody who uses one. The naming is right and the structure is right, and that is not the same as the page being pleasant to hear. This is the gap we would most like closed and we will pay somebody to close it.
- No testing with voice control, a switch device, a braille display or magnification software.
- The client portal is behind a sign-in and has not been through the same rendered pass as the public pages. It is held to the same standard and has not yet been measured against it. If you are a client and something in the portal does not work, that is the first thing we want to hear about.
- The audit findings your engagement produces include charts. They carry their figures as text as well, but they have not been checked as thoroughly as these pages.
- PDFs we send, including the engagement agreement, have not been checked for accessibility tagging.
If something does not work
Write to us through the contact form, or reply to any thread you already have with us, and say what you were trying to do and what got in the way. You do not need to know the name of the problem or which guideline it is.
We will reply within five business days, tell you whether we can fix it and when, and if a fix is going to take a while we will offer you the same information another way in the meantime, which for a client means a person reading it to you on a call.
We are a two-person firm and this is not a queue. It is one of us reading it.
If something on this page contradicts an agreement you have signed with us, the agreement governs. Tell us, because one of the two is wrong and we want to know which.
Ask us about it