
WebAIM, a group that studies web accessibility, runs an automated scan of the top million home pages each year. The 2026 results found errors on 95.9% of them, averaging 56 detectable barriers per page. That is worse than the year before, reversing six years of gradual improvement.
The useful part is not the headline number. It is that the six most common failures have been the same for seven years running, and none of them are obscure. Low contrast text is the biggest one by far, showing up on 83.9% of pages at roughly 34 instances each. That is text that is too pale against its background to read comfortably. Then there are form fields with no proper label attached (about a third of all inputs on an average home page), headings used out of order so the page structure does not match what you see on screen (41.8% of pages), and pages using more than one top-level heading (18.1%).
For your site, this matters in three practical ways. First, these are the same things that make a page hard to read for anyone squinting at a phone in Arizona sun, not just for people using a screen reader. Second, heading structure is what search engines use to work out what a page is about, so getting it wrong costs you twice. Third, form labels are the difference between a contact form people complete and one they abandon.
Also worth knowing: WCAG and Section 508 are the two standards that come up in this conversation. WCAG is the international guideline set. Section 508 is the US federal rule, which for most of these items just points at WCAG anyway. They are not a separate discipline. They are a written-down version of things a competent designer already does.
What to do: pick your busiest page and check two things yourself. Can you read every bit of text easily, including grey text on white and text sitting over a photo? And does every field on your contact form have a visible label above or beside it, not just faded placeholder text inside the box? If either fails, that is a short, cheap conversation with whoever maintains your site.
Reported by:


