Accessibility overlay or real fixes — what to choose for your site

You launch a site and an email arrives: "one line of script and your site meets accessibility standards." The vendor's page looks convincing — a little person icon in the corner, font size sliders, a colour-blind mode.
Tempting. Except the people it's supposedly for install browser extensions to block those widgets.
That's not a figure of speech — it's what screen reader users say in the quotes collected on the industry fact sheet about overlays. Let's unpack how it got here and what to do instead.
What an overlay is and what it promises
An overlay is third-party JavaScript you add in one line, running in the visitor's browser. It usually bundles two different things:
- A widget — a settings panel: bigger text, higher contrast, highlighted links, a "blind mode."
- Automated repair — a script that scans the page on load and tries to add missing labels, alt text and roles by itself.
The promise is "compliance without code changes and without a developer." That promise is where it breaks down.
Five criteria where they diverge
1. What gets fixed. An overlay patches the page in the browser, on top of finished markup. Code fixes address the cause: a <div> becomes a <button>, a field gains a <label>. Net result: the overlay masks a symptom, the code removes the error.
2. Reliability. The industry fact sheet — signed by editors of the WCAG, ARIA and HTML specifications, contributors to the JAWS and NVDA screen readers and accessibility specialists from Google, Microsoft, Apple and the BBC — explicitly lists what automation does not do reliably: applying text alternatives to images, repairing field labels and error handling, focus control, keyboard access. That list is essentially what accessibility is made of.
3. What users think. In the WebAIM survey of accessibility practitioners, 67% rate these tools as not at all or not very effective. Among respondents with disabilities it's 72%, with only 2.4% calling them very effective.
4. Legal risk. The vendors' main selling point. The fact sheet is blunt: no overlay product on the market can make a website fully compliant with any existing accessibility standard, and therefore cannot eliminate legal risk.
5. Price and horizon. An overlay is a subscription you pay for as long as the widget is installed: stop paying, you're back to zero. Code fixes are done once and stay in the project. For a small site that's usually an evening: real tags, labels on fields, contrast, heading order.
Why the widget gets in the way
Here's the least obvious part. Surely an extra settings panel can't hurt?
The logic goes like this: if a person genuinely needed those features, they'd need them on every site on the internet. So text enlargement, high contrast and speech are already configured — in their operating system, browser or their own software. At best, the widget duplicates what they already have.
At worst it conflicts. From real complaints: focus starts jumping around the page; everything becomes a heading, including table cells with prices, turning the outline into mush; a "4 stars" rating gets announced as "unpronounceable." One more detail: some overlays try to detect that a visitor is using a screen reader — and people disable that detection in their browser to get a usable site back.
Who should pick what
The verdict, without hedging.
Fix the code — almost always. Especially if the site is yours or was built by vibe coding: you have access to the markup, and fixing the cause is cheaper than paying a subscription to mask it. Start with a screen reader pass — you'll get five concrete tasks instead of an abstract score.
An overlay is arguably justified in one case: a legacy system whose code you physically cannot touch, where some repair is needed by tomorrow. And even then, as a temporary crutch while it's being rewritten, with a clear-eyed understanding that this is not compliance.
Installing the slider widget is never necessary. It's pure duplication of the user's own system settings, and sometimes an active obstacle.
Is it true that a widget "makes my site legally compliant"?
No. Automated repair closes some formal errors, but not all of them, and the promise of full compliance from one line of script isn't backed by experts or by how lawsuits have actually gone.
How long do manual fixes take?
For a small site, usually an evening. The core work: real <button> elements instead of clickable blocks, <label> on fields, alt text, 4.5:1 contrast, and a visible focus ring.
So automated checkers are useless?
Not at all — but as a way of finding problems, not fixing them. Run axe or Lighthouse, take the list, fix it in code: that works. Trouble starts when a script edits the page on your behalf.
The AI Architect 2027 course: step by step from computer basics to your own app and first customers. 543 short lessons on your phone.





