Why accessibility overlays don't make you compliant, and can make you a bigger target
If someone has already sold your practice a one-line accessibility "widget," this is worth five minutes before you assume the box is checked. It probably isn't, and the widget itself has become part of the problem it claims to solve.
What an overlay actually is
An accessibility overlay is a single line of JavaScript you add to your site. It sits on top of your existing pages and tries to patch accessibility problems at runtime: adjusting contrast, adding labels it guesses at, and intercepting keyboard input. The pitch is simple: instant compliance, no developer time, one script tag.
The problem is what "on top of" means. An overlay doesn't touch your actual HTML. It layers a best-effort patch over markup that's still broken underneath, using automated heuristics that get it wrong often enough to matter.
Why that doesn't hold up
What an overlay does
Guesses at fixes using pattern-matching and machine learning, applied after the page has already loaded. It cannot know what an image actually depicts, whether a form field's purpose is genuinely conveyed, or whether your specific booking flow makes sense to someone using a screen reader.
What actually fixes it
Correcting the underlying HTML, alt text, and ARIA attributes so that any assistive technology, not just the one the overlay vendor tested against, renders the page correctly, the same way a properly built page would from the start.
This isn't a fringe opinion. It's the position of the accessibility testing community broadly, and it's the reason the standard itself (WCAG 2.1 AA, the one the HHS Section 504 rule names directly) is written in terms of how content is built, not what runs on top of it after the fact.
The part that should actually change your mind
Installing an overlay hasn't kept sites out of court
Installing an overlay doesn't just fail to fix the underlying problem. It doesn't reliably prevent the lawsuit either. UsableNet, an accessibility vendor that tracks these filings, reported that as of mid-year 2024 over 20% of digital accessibility lawsuits were filed against companies using accessibility widgets.1 And promising instant compliance is not the same as delivering it. In April 2025 the FTC finalized a $1 million order against the overlay vendor accessiBe, settling allegations that its claims a product could make any website WCAG-compliant were false, misleading, or unsubstantiated.2
What to do instead
- Find out what's actually wrong. A real WCAG 2.1 AA scan tells you concretely: not a guess, not a patch, an actual list of what fails and where.
- Fix the markup. Most findings are genuinely an afternoon's work for whoever maintains the site once someone tells them which ones matter.
- If you already have an overlay installed, don't assume it's covering you. Run a scan of your site as it actually renders. The gaps the overlay didn't catch are exactly the evidence worth knowing about before anyone else finds them.
Sources
- UsableNet, “Why Your Website's Accessibility Widget Might Be a Lawsuit Waiting to Happen”. This is UsableNet's own count; UsableNet also sells accessibility services.
- Federal Trade Commission, “FTC Approves Final Order Requiring accessiBe to Pay $1 Million”, April 22, 2025.
See what an overlay isn't catching
Free WCAG 2.1 AA scan of your homepage plus your four highest-risk pages, including sites that already have an overlay installed. Report emailed as soon as the scan finishes.
Scan my website freeQuestions
Doesn't an overlay at least show good-faith effort?
It can, but good-faith effort and actual conformance are different things, and only one of them is what the standard measures. A dated record of real fixes is stronger evidence of good faith than a script tag.
We can't afford to rebuild the whole site right now.
You don't need to. Most WCAG 2.1 AA findings are targeted fixes: specific elements, specific pages, not a rebuild. A prioritized list ranked by legal exposure lets you fix the highest-risk items first without touching everything at once.
Is any automated tool actually trustworthy, then?
An automated scan and an automated overlay solve different problems. A scan tells you what's wrong so a person can fix it. An overlay tries to hide what's wrong without a person involved at all. The first is honest about its limits; the second is sold as a replacement for fixing anything.