What does an accessibility findings report actually look like?
Real findings, from a real test — run in public, on a site built to be tested.
Clients often ask what they will actually receive at the end of an audit. This page answers that honestly: it is a short findings report, written the way I write them for clients, from a test anyone can repeat. The test site is the “Before and After Demonstration” home page published by the W3C Web Accessibility Initiative — the international body that writes the web accessibility guidelines. It is a fictional news site the W3C maintains deliberately in an inaccessible “before” version, so people can learn what barriers look like. No client’s site is shown, and nothing here is invented: every finding below is in that page’s code today.
Finding 1 — Critical: the page throws keyboard focus away
What happens: the top story links contain code that removes focus the moment a keyboard user tabs to them. Pressing Tab can never land on the day’s main stories — the page itself pushes the keyboard away.
Who it shuts out: everyone who navigates by keyboard — including blind screen-reader users and people with motor disabilities.
The fix: remove the focus-removing behaviour entirely, and make sure every link shows a clearly visible focus outline. Tested by completing the whole journey keyboard-only.
Finding 2 — Critical: images doing jobs, with no text alternative
What happens: the navigation images and headline images carry no alternative text at all — a screen reader announces a file name, or nothing useful. One decorative image’s alternative text is literally the word “bullet”, and the organisation’s phone number exists only inside an image.
Who it shuts out: blind users, who hear noise where sighted users see the menu, the headlines and the phone number.
The fix: real text for anything that informs or navigates; empty alternatives for purely decorative images; and contact details as text, not pictures.
Finding 3 — Serious: headlines that aren’t headings
What happens: the news headlines are styled paragraphs — big font, no heading markup. The whole page contains a single heading element, so the structure a screen reader offers for jumping between stories simply doesn’t exist.
Who it shuts out: screen-reader users, who navigate long pages by headings the way sighted users scan with their eyes.
The fix: mark headlines up as real headings, in order — one page title, section headings, then story headings beneath them.
Finding 4 — Serious: interactions that only answer to a mouse
What happens: the arrow graphics change only when a mouse hovers over them. Keyboard and touch users get no equivalent response — the page treats the mouse as the only way in.
Who it shuts out: keyboard, switch-device and many touch users.
The fix: build interactions as real links and buttons that respond to focus and activation from any input — not hover alone.
Finding 5 — Moderate: the page never says what language it is in
What happens: the page’s code declares no language, so a screen reader may read English text with the wrong pronunciation rules.
Who it shuts out: screen-reader users — a small defect that degrades every single sentence.
The fix: declare the language in the page code — a one-line repair.
In a real project
A client report covers your real journeys — buying, booking, applying, submitting — tested manually with JAWS, NVDA, VoiceOver and TalkBack, with every finding prioritised like the ones above, a plain-language explanation your developers can act on, and a walkthrough call to go through it together. That is what the AODA Check and full audits deliver.