Web accessibility remediation (WCAG / ADA / EAA)

What this is

I went looking for a problem shaped like a retainer — something a company has to keep fixing, that a tool can't fully automate, where going deep beats going wide. DMARC failed that test on every axis. Web accessibility passes it on every axis. This is the research that convinced me, and the questions I'm still holding open.

If you run an e-commerce site and someone sold you an accessibility widget, the two-minute version is: it is not doing what you were told it does. Read on for why, and what actually protects you.

---

The problem

The legal baseline for "accessible" is WCAG — the Web Content Accessibility Guidelines. In practice it means your site works for someone who can't see it (a screen reader announces every image, button, and form field) and for someone who can't use a mouse (everything is reachable and operable by keyboard). Most sites fail this badly: the annual WebAIM survey of the top million home pages finds ~96% have detectable WCAG failures, averaging dozens per page.

That used to be a moral argument. Now it's an enforcement one, on two continents at once:

• United States — the ADA. Federal courts took roughly 3,100 web-accessibility cases in 2025, up ~27% year over year, and about 70% target e-commerce. A demand letter arrives, the number to make it go away is five figures, and the cheapest defense is to already be compliant.
• European Union — the EAA. The European Accessibility Act came into force on 28 June 2025. It reaches any business selling to EU consumers, not just EU-registered ones, and the first enforcement is already landing on retailers under daily fines. For a store with EU customers, non-compliance is now a market-access problem, not a lawsuit-someday problem.

So the demand is real, it's dated, and it's growing. The interesting part is what companies did about it.

---

The trap most companies are already in

Faced with that, the obvious move is to buy an overlay — a single line of JavaScript (accessiBe, UserWay, AudioEye and others) that promises to make a site "ADA compliant" automatically. Millions of sites run one. Here's the problem:

• It doesn't work, and that's now official. In 2025 the FTC settled with accessiBe for roughly $1M over marketing an overlay as delivering compliance it did not deliver. The tool cannot fix the underlying HTML; it papers over it at runtime, and assistive-technology users routinely report it makes the site harder to use.
• It's a target, not a shield. Analyses of ADA web suits find roughly one in five defendants already had an overlay installed. Plaintiffs' attorneys specifically look for the widget — its presence signals a company that believes it's covered and hasn't actually been audited.

That last point is why an overlay-carrying store is, counter-intuitively, the best-qualified lead in this whole space: it has budget (it already paid for the widget), awareness (it knows accessibility is a liability), and a false sense of safety that a two-minute demonstration punctures.

---

Why a tool can't just solve this

If software could fully audit and repair accessibility, this would be a SaaS play and I'd be too late. It can't, and the ceiling is well documented:

• Automated scanners catch 25–50% of WCAG issues. They reliably find missing alt text or an absent lang attribute. They cannot judge whether alt text is meaningful, whether a custom widget's keyboard behavior makes sense, or whether a screen-reader user can actually complete checkout. Those require a human running the flow.
• The regulator said so. In its rulemaking the DOJ was explicit that automated tools and AI cannot, on their own, make a site compliant — human testing and remediation remain necessary.

That gap — the half to three-quarters of the problem no tool touches — is the whole opportunity. It's a knowledge-and-judgment moat, and it's labor that recurs every time the site changes. A standard you must meet, that automation can't finish, on a codebase that keeps moving: that is the definition of a retainer, not a one-off.

---

Where this leaves me

DMARC was a real global market reachable by nobody I am, priced at the floor, and the wrong shape. Accessibility is a real, dated, growing market; the warmest segment (overlay-carriers) is passively identifiable from the public page; the moat is judgment a tool can't replicate; and the work regenerates. It clears every bar DMARC failed.

What I don't yet know is entirely about the channel — reply rates, the law-firm referral question, and whether the mid-market pays the bands the market quotes. Those are the four questions above, and they're the ones I'm spending the next stretch answering. If you've sold remediation, or bought it, I'd trade notes.

Meanwhile, the thing under the research is a service I can already deliver: if you're running an overlay and want to know what it's not covering, that's a real conversation.