A report extract from our own website audit
Making enquiry-form errors easier to read
See the barrier, the evidence, the recommended correction and what the retest established.
The task and barrier
A visitor needs to correct missing or invalid information before sending an enquiry. On the version of our contact page captured on 7 October 2026, the red field labels and error messages were too faint against the white form background. This could make the instructions harder to read for someone with low vision.
Finding AAT-001: insufficient text contrast. We assigned medium priority: the instructions remained available, but their presentation made correcting the form harder.
Evidence and requirement
The earlier capture contained four invalid field labels and four error messages for Name, Email, Subject and Message. These were eight occurrences of one finding. This normal-sized text required at least 4.5:1 contrast under WCAG 2.2 success criterion 1.4.3 Contrast Minimum.
Earlier capture on 7 October
Light red #ef4444 on white measured 3.76:1, below the requirement. Error text was 14 CSS pixels, with font weight 500.
Preview retest on 30 October
Dark red #981b1b on white measured 8.37:1, above the requirement in the retested states. Field errors were 14 CSS pixels with weight 700; the enquiry-choice error used weight 500.
How the issue was reproduced
- Open the contact page and leave the required fields empty.
- Activate the submission button to display the validation messages.
- Identify the visible error labels and messages, and measure their computed text and background colours.
- Calculate the contrast ratio and compare it with the requirement for the text size.
The earlier test blocked outgoing submissions. The retest used an isolated preview with a test-only enquiry form. No real enquiry was sent.
Recommended correction
Use a darker colour for error messages and invalid labels, checking it against the background where the text actually appears. Check the error-summary links too. Keep clear error wording so the visitor can understand what needs correcting.

What the retest established
On 30 October, empty-form and invalid training-enquiry states were tested at 1440 × 1000 and 390 × 844 CSS pixels. The red error text, invalid labels and error-summary links measured 8.37:1 against white in these four states. The error-summary heading and other measured labels had higher contrast.
Retest result: the measured text meets the contrast requirement in the tested states.
Why manual screen-reader testing also matters
A visible error is only part of the story. A sighted visitor may notice a red outline, while a screen-reader user needs a clear error message associated with the field and a usable way to reach and correct it.
Manual testing checks what is spoken, where keyboard focus goes and whether the person can recover from an error. Founder Aj Ahmed's everyday JAWS use informs those checks. Read more about our screen-reader testing experience.
Scope of this extract
This extract covers one type of barrier. It compares a historical live-site capture with an identified development preview. The form changed between those dates, so the retest covers the current fields rather than an identical set of eight text occurrences.
Testing used Chromium on Windows, browser interaction and computed-style measurements. The narrow check used a resized desktop browser. Screen-reader speech, other browsers, dark or forced-colour modes, physical phones and the complete enquiry process were outside this contrast retest. The result does not certify the whole website.
Technical evidence record
Earlier capture: 7 October 2026, Chromium 149.0.7827.55, 1440 × 1000. Retained HTML and asset hashes identify the captured deployment; its exact Git commit is not established.
Retest: 30 October 2026, Chromium 156.0.8078.79, Netlify deployment 6ae4f03d102b626149824690, revision b8247e654434d8f15fff8f40ed3a4ba7358d8306.
Unrounded ratios were 3.763127680148142:1 before and 8.374275666573721:1 for retested red text. Decisions used unrounded values. The retest recorded 128 text-element observations across four states; these are not separate findings.
We retain the original screenshots, computed styles, tested markup and review history. This extract compares the observed implementations; it does not attribute the historical fix to a particular source-code change.
Get findings your team can act on
We agree the pages, user journeys and testing environments before an audit. You receive a prioritised written report with step-by-step guidance. A call is available on request, and follow-up and retesting are agreed in your quote.
Request an audit quote