You have probably seen the button. A small figure with its arms out, parked in the corner of the screen, and behind it a panel of switches - bigger text, more contrast, pause animations, highlight links, a screen reader. It costs about what a phone contract costs, and the sales page says it will make your website compliant. Somebody sent me one this week and asked whether we should put it on their site. The short answer is no. The longer answer is worth writing down, because the reasons are not obvious from the outside and the pitch is a genuinely good one.
What an overlay actually is
An accessibility overlay - also sold as a widget, a toolbar or an accessibility plugin - is one line of somebody else's JavaScript that you paste into your website. When a visitor opens a page, that script downloads, runs inside their browser, and starts editing the page that has already arrived.
It does two jobs. The visible one is the panel: a set of controls that change how the page looks on that visitor's screen. The invisible one is a set of automatic corrections - the script inspects the page, guesses at things it thinks are wrong, and patches them on the fly. Missing form labels, missing image descriptions, unclear link text.
Both jobs happen in the visitor's browser, after the page has loaded, every single time. That one fact is the whole story, and everything below follows from it.
What it does to your website
Nothing. That is not a rhetorical flourish - it is a description of the mechanism. The script never touches what your website sends out. It edits a copy in the visitor's browser. Turn JavaScript off, block the script with an ad blocker, hit the page while the vendor's servers are having a bad afternoon, and every barrier that was there before is there again, because none of them ever went anywhere.
The clearest example is image descriptions, because it is the one place where the limits are impossible to argue with. Say you have a photograph on a product page. A blind visitor's screen reader needs a short line of text describing what it shows. If nobody wrote that line, no script can invent it - because the useful description is not a description of the pixels, it is a description of why the picture is on that page. Automated tools have got startlingly good at telling you there is a yellow machine on gravel. They cannot tell you it is the 80-tonne model your customer is comparing against the 60. Best case, the visitor gets something vague. Worse case, they get something confidently wrong, which is worse than silence, because now they think they know.
The same goes for the rest of it. Heading structure that runs h1 to h4 with nothing in between is still doing that. A form field whose label was never joined to the input is still unlabelled in your code. A contrast failure is in your stylesheet, where you can fix it in one line, and the overlay is repainting it in the browser instead, for the fraction of visitors who find the panel and switch it on.
An overlay changes the picture on the screen. It does not change the website. Anything it appears to fix is unfixed the moment the script does not run.
The switches your browser already has
Look at what is actually in the panel. Bigger text. More contrast. Larger cursor. Pause animations. Wider letter spacing. Change the line height.
Every one of those already exists, one layer down, and works better there. Text size is Ctrl and plus, or Cmd and plus, in every browser ever made. Contrast, cursor size and reduced motion are settings in Windows, macOS, iOS and Android - and a properly built website already listens for the reduced-motion one and stops moving without being asked. Reading and spacing tools are built into Safari, Edge and Firefox, and there are good free extensions for the rest.
Here is the difference that matters. Somebody who needs larger text sets it once, in their operating system, and it applies to every website, every app and every email for the rest of their life. A widget sets it on your site only, remembers nothing, and has to be hunted for and reconfigured on the next site - if that site happens to have bought the same product, which it probably has not. You are asking the person with the most friction in their day to do more work, for a smaller result.
The screen reader button is the strangest one of all. Anybody who needs a screen reader has a screen reader. It runs at the operating-system level, it reads their email and their banking app as well as your website, and it is tuned to a speech rate that would sound like a fax machine to you. Offering them a different one, that works on one website, is not an accommodation. It is a stranger handing you a spare pair of glasses.
Where it makes things worse
If overlays were simply useless, they would be a waste of a subscription and nothing more. The reason accessibility practitioners are so consistently against them is that a script rewriting the page can collide with the software the visitor is already running.
Screen readers, magnifiers and switch devices read the page structure as it is delivered and build a model of it. An overlay that reorders focus, rewrites roles or injects its own controls can break that model - and the failure is not obvious to whoever installed it, because it happens in a piece of software they do not use. There are well-documented cases of overlays breaking things on a page that were working correctly before they arrived. The panel itself is one more thing to tab past on every page, and it is fixed in a corner where it can sit on top of your own content at high zoom.
There is a plainer commercial harm too. An organisation installs the widget, ticks accessibility off the list, and stops. The budget that would have fixed the actual site now goes on an annual subscription, and the site quietly stays exactly as inaccessible as it was, minus a few hundred pounds a year. That is the real cost, and it is much larger than the licence fee.
This is not a fringe view. The Overlay Fact Sheet, an open statement against the use of these products, has been signed by hundreds of accessibility practitioners - including people who wrote and edit the WCAG, ARIA and HTML specifications themselves, and internal accessibility specialists at some of the largest technology companies in the world. When the people who wrote the standard tell you a product does not meet the standard, that is worth more than a testimonial.
Where UK law actually leaves you
This is the part the sales page is quietest about, so let us be precise - and let us be careful, because most of what is written about overlays online is American and the UK position genuinely is different.
Under the Equality Act 2010, a service provider has to make reasonable adjustments so that disabled people are not put at a substantial disadvantage. The test is about the barrier. If somebody cannot complete your booking form with a keyboard, a widget that lets them enlarge the text has not removed that barrier - it has decorated it. Nothing you buy discharges a duty that is measured by whether the person could actually use the service.
Under the Public Sector Bodies Accessibility Regulations 2018, it is even simpler. The Government Digital Service monitors a sample of public sector sites and tests them - it monitored over 1,200 websites between January 2022 and September 2024 alone. The testing looks at the site. And the accessibility statement you are required to publish has to say honestly what does not conform; an overlay does not change a single word of what you have to disclose, because it has not changed a single thing about the site.
What about the American material? The US figures are striking - in the first half of 2025, over a fifth of American web accessibility lawsuits were filed against sites that already had an overlay installed - and in 2025 the US Federal Trade Commission imposed a one million dollar penalty on one of the largest overlay vendors over the claims made in its advertising and over undisclosed paid reviews. But the UK has nothing resembling that litigation industry, so I am not going to tell you that you will be sued, because in Norfolk you very probably will not be. The honest UK argument is duller and stronger: you would be paying an annual fee for something that does not do the thing it is being bought to do.
What to do instead
Fix the site. That is the entire answer, and it is less daunting than it sounds - most sites have a handful of repeated problems rather than hundreds of unique ones, because the same header, the same form and the same card component appear on every page. Fix the component and you fix it everywhere.
If you want a first read on where your own site stands, our guide to accessible web design has five checks you can run yourself this afternoon, free, without a specialist. The keyboard one takes about ninety seconds and tells you more than any scan.
One fair distinction, because this is where the argument gets lazy. A control that belongs to your site is a different animal from an injected overlay. A proper light, dark and high-contrast theme, built as part of the site, changing real styles that you wrote and tested, is a genuine feature - we build them, and people use them. So are reading tools for people who want them. The difference is that these are part of the thing, tested with it, and they make no claim about conformance. They sit on top of an accessible site. They are not sold as a substitute for one.
What we are not claiming
We are not claiming everyone selling these is acting in bad faith. Plenty of web designers install them believing the marketing, because the marketing is aimed squarely at people who know they have an obligation and do not have the budget to meet it properly. That is a sympathetic position to be in. It is still the wrong purchase.
We are not claiming that a site without a widget is accessible. The absence of an overlay proves nothing at all. Most sites that have never heard of one are also failing, quietly, in the same ways.
We are not claiming our own site is perfect. We build to WCAG 2.2 AA and we test against it, but the assessment of neooptic.com is our own - we say so plainly in our accessibility statement, which also declares that no third party has audited it. We would rather tell you that than imply a certificate we do not hold.
And we are not claiming nobody gets anything from the panel. Some visitors do use the contrast and text-size controls, and if that is genuinely helping somebody on your site today then that is a real thing and it counts. The objection is not to a toolbar existing. It is to a toolbar being sold as compliance.
The question worth asking
If somebody offers you one - a vendor, a reseller, or the agency that built your site - two questions will settle it quickly.
Which specific problems does this fix in the site itself, so that they stay fixed if the script never loads? An honest answer is "none, it works in the browser". Any other answer is worth probing.
And will you put in writing that installing this makes us compliant? Watch what happens. The marketing says it; the contract, almost always, does not.
If you are in Norfolk and want a straight read on where your site actually stands, ask us - the five free checks first, and an honest answer about whether you have a tidy-up or a rebuild. We would rather you spent the widget money on the site.
