The pitch is close to irresistible. One line of JavaScript, no engineering work,
no audit, and a legal problem becomes a subscription line item. If it worked, it
would be the best value in software.
There is a lot of argument about whether it works. There is less argument among
the people it is supposedly for, and that is the part worth starting with.
One number
The WebAIM survey of web accessibility practitioners asked how effective overlays
are. Of 758 valid responses, 67% rated them "not at all" or "not very"
effective.
Among respondents who themselves have disabilities, that rose to 72%.
Only 2.4% of them called overlays very effective.
That survey is from January 2021, so treat the exact figures as of their date. I
am not aware of a later practitioner survey that reverses the finding, and if one
exists I would genuinely like to see it.
The shape of the result is the part that matters: the group with the most direct
experience of using these tools rates them worse than the general population
of practitioners does.
For any other product that finding would end the discussion. A wheelchair ramp
that wheelchair users rate worse than non-users do is not a ramp.
What the FTC order actually says
In January 2025 the US Federal Trade Commission brought a case against accessiBe
and approved the final order in April.
One million dollars.
The interesting part is not the money. It is the prohibition. The order bars
accessiBe from representing that its automated products can make any website
WCAG-compliant, or can ensure continued compliance over time, unless it has
evidence to support the claim.
Read that as a statement about the category rather than about one company. A
regulator looked at "automated tool makes your site compliant" and concluded it
could not be said without evidence, evidence that, in that case, did not exist.
The same complaint also covered third-party articles and reviews formatted to look
like independent opinions, where the company's connection to the reviewer was not
disclosed. If you evaluated overlays by reading reviews a few years ago, some of
what you read was not what it appeared to be.
Why it cannot work, mechanically
An overlay is JavaScript that runs in the visitor's browser and modifies the page
after it arrives. That single fact produces most of the limitations.
It has to guess. No script can know that an image of a chart is decorative in
one place and load-bearing in another, or that your heading structure is wrong
rather than unusual, or which of four inputs is the one labelled "Name" in the
visual layout. Automated remediation guesses, and a confident wrong label is worse
than a missing one: a missing label is a barrier somebody can work around, and a
wrong label sends them somewhere else entirely.
It fixes nothing at the source. The changes exist in the rendered page while
the script is running. They are not in your markup, not in your templates, and not
in the next page your team ships. Turn the subscription off and every fix leaves
with it.
It depends on somebody else's infrastructure. The script is served from a
vendor CDN. If that CDN is slow, blocked by a corporate proxy, or down, the
remediation is simply absent and absent precisely for the users least able to
compensate.
Some things are structural. Focus order, keyboard traps, whether a multi-step
checkout can be completed without a mouse, whether error messages are announced at
all. Those are properties of how the application is built. A layer painted over
the top does not change them.
The litigation data
According to UsableNet's 2026 midyear report,
about 20% of the companies sued over digital accessibility in the US already
had an accessibility widget or overlay installed.
That number does not prove overlays cause lawsuits. Sites that install one are
plausibly sites already worried about being sued, which is a selection effect and
should be said out loud rather than glossed over.
What it does refute is the marketing claim. Roughly one defendant in five had
already bought protection. Whatever the widget is, it is not a
shield and its presence is trivially detectable by anybody looking for
defendants.
Where the overlay vendors are not wrong
A fair argument has to include this part.
User preference controls are genuinely useful. Being able to raise text size,
increase contrast or reduce motion helps real people, and some users like having
those controls in one place rather than hunting through browser settings. That is
a legitimate feature and it is a pity it arrives bundled with a legal claim.
Some organisations have nothing at all. For a site with no accessibility work
whatsoever, a widget that adds a few keyboard shortcuts is not worse than nothing
in every respect.
The underlying problem is real and hard. Vendors did not invent the demand.
Companies genuinely are being asked to fix years of accumulated markup with no
budget and no expertise, and "one line of JavaScript" answers a real question,
just not truthfully.
The objection is not that the software exists. It is the compliance claim attached
to it, and that claim is now the subject of a regulator's order.
What to do instead, in order
Find out where you actually stand. Not a score out of a hundred: a list of
which clauses of the harmonised standard are failing, and how badly.
npx curbcut https://example.com --crawl
Or check a page in your browser with nothing
installed.
Fix the task-blockers first. An unlabelled password field in your checkout
outranks four hundred low-contrast footer links, however your tool sorts them by
default.
Get a person to test what automation cannot reach. Automated testing gets you
roughly a third of the way. The rest: keyboard journeys, focus order, whether alt
text says anything useful needs a human, and some of it needs a human who uses
assistive technology every day.
Write a statement you can defend. The template most people copy is for a
different law,
and an unsupported conformance claim is the first document anybody investigating
you will read.
The part where I declare my interest
I build an accessibility tool, so read all of the above knowing that.
Here is what mine does not do: it does not fix anything. It finds what automation
can find, maps it onto EN 301 549 clauses, ranks it by regulatory exposure, and
prints the clauses it could not evaluate rather than staying quiet about them. The
fixing is engineering work somebody has to do, and no product I could build would
change that.
More than a thousand accessibility practitioners including contributors to the WCAG,
ARIA and HTML specifications, and accessibility staff at Google, Microsoft, Apple,
Shopify, eBay and Target have signed the
Overlay Fact Sheet, which sets out the
technical case in more detail than I have here. Read it before you buy anything,
including mine.
Top comments (0)