Accessibility
WCAG failures, ranked by what they stop you doing.
Every page is audited daily and again within minutes of a deployment, because accessibility regressions ship with markup changes rather than with traffic.
An automated pass covers about a third of WCAG.
That third is worth automating: it is the part that breaks silently during a redesign. The other two thirds need a person, and Tenvey says so on the page instead of showing a green tick for rules it never ran.
Checked on every run
- Contrast of every text node against the background it is actually painted on.
- Images with no alt attribute, and alt text that only repeats the file name.
- Form controls with no label associated with them.
- Landmark structure: one main, headings in order, no jump from h2 to h4.
- Interactive elements under 44 × 44 px, measured on a touch viewport.
- ARIA that contradicts its element — role="button" on a link that navigates.
- Focus rings removed without a replacement, and tab order that fights the DOM.
Not checked, and never claimed
- Whether the alt text describes what the image actually shows.
- Whether a label says what the field is for, or merely exists.
- Whether a keyboard path through a whole flow still makes sense at the end.
- Whether an error message tells someone using a screen reader what to change.
These four are in the report as an unchecked list, with the pages they apply to. An unanswered question is more useful than a green tick that answered a different one.
A ratio, against the line it has to clear.
They run before anything is painted and hold the page back by 640 ms on a mid-range phone.
Accessibility scores 91 and carries 20% of the health score.
Contrast · homepage · 3 instances
InfoThree scripts block the first render
2.81:1 measured
4.5:1 required for AA
analytics.js 34 KB · chat-widget.js 128 KB · slider.min.js 61 KB · all render-blocking in <head>
Next step
Move all three to the end of the body or load them with defer. None is needed for the first paint.
Three questions decide where it lands.
Severity labels from a rule set are the same on every site. Ranking has to know something about yours.
- Where it sits
- The same contrast failure is an info finding on a newsletter button and a critical one on the checkout submit. Tenvey knows which paths you marked as critical.
- How many times
- 3 instances, all of them the same component. One change closes all four, so it is ranked as one issue and not as four.
- Blocks or slows
- A field with no label is unusable with a screen reader. Low contrast makes a label harder to read. Both are reported; only the first is ever ranked above a broken checkout.
Also watched