Accessibility Statement

What we are building towards

We build this site against WCAG 2.2 Level AA, an international standard for making websites usable by people with disabilities. That is the target we write and test against. It is not a claim that we have reached it. This page says what we actually check, what those checks miss, and examples of what we already know is wrong.

What we check

We run an automated accessibility checker over the pages this site builds. It tests against the WCAG 2.2 Level AA rules, plus a set of rules about page structure: that headings follow a sensible order, that each page has one main region, that the parts of the page are named and distinct, and that the link which skips past the navigation points at something real.

Alongside that, a small number of checks are written by hand. They cover whether keyboard focus moves into the page content after you follow a link, whether the controls we test can be opened and operated from the keyboard alone, and whether text and form fields stay readable on a narrow screen.

These checks were last run on 15 September 2026.

What these checks do not cover

An automated checker finds only a fraction of real accessibility problems. It can see that a label is missing. It cannot see that a label is present and makes no sense, that the reading order is confusing, or that a task cannot be finished without a mouse. Those are only found by a person using the site. Nobody has yet worked through this site with a screen reader.

The checker also reports items it cannot decide, and there are a great many of them. Most are text sitting over a moving background, where contrast cannot be measured. The rest are a repeated identifier, an attribute used where it is not allowed, and links inside blocks of text. We record all of these and leave them open. We do not count them as passing.

There is one more gap, and it is specific to the page you are reading. Most of our checking is done against an internal build of the site, which contains more pages than the public one, and the findings below were measured there. The public build is scanned too, the home page included. Some of the hand-written checks do not run on it, because the pages and controls they exercise are not published yet.

A passing check means a known problem has not come back. It is not evidence that the page works for you.

Some of what we know is wrong

These are examples, not a complete list, and they are what our own automated checks reported on our internal build when they were last run. Some of them are on pages that are not published yet. Problems no automated checker can see are not here at all, and neither is anything a person using assistive technology would find, because nobody has done that pass.

• Text contrast was below the required level on the report, its methodology page, and two of the pages about us. We raised it on 15 September 2026 and those checks now pass on every page this site builds. That covers only the contrast a checker can measure, which is the next point but one.
• On one page, two parts of the layout are not named distinctly enough to tell apart, which matters if you navigate by landmark.
• A large amount of the contrast on our heavier pages, the home page among them, cannot be measured automatically at all. Those parts are unverified rather than known to be correct.

Telling us about a barrier

If something here stops you from doing what you came to do, please write to us and say what you were trying to do and what happened: hello at flowdealer.ai