The Accessibility Mistakes Even Big Brands Still Make

You’d think accessibility failures would correlate with budget. Small sites built by one overworked founder should be full of them; giant brands with design systems, dedicated platform teams, and legal departments should be clean. Spend an afternoon with a screen reader on famous homepages and you’ll find it isn’t so. The same handful of mistakes shows up on nine-figure websites, and I think the reasons are more interesting than the mistakes.

Start with the mistakes themselves, because they’re remarkably consistent. Focus outlines removed, so keyboard users can’t see where they are; somewhere in the stylesheet there’s a rule stripping the outline because a designer found the default ring ugly, and no one added anything in its place. Carousels that auto-advance, whisking content away mid-read, with pause buttons that are tiny or absent and slide dots a screen reader announces as “button, button, button.” Interactive elements built from divs and spans, clickable to a mouse and nonexistent to a keyboard. Mega-menus that open on hover and collapse the instant a trembling hand drifts a few pixels off course. Video that autoplays. Text baked into promotional images, invisible to assistive technology and unenlargeable by anyone. None of this is exotic. It’s the same short list, year after year, on sites that plainly cost millions.

So why doesn’t money fix it? The first reason is that accessibility isn’t a feature you buy once; it’s a property you maintain, and big sites change constantly. Many large companies did the audit, did the remediation, got clean, and then kept shipping. Every campaign microsite, every redesign, every third-party marketing tag is a chance to regress, and a large organization takes more of those chances per week than a small one takes per year. Accessibility debt isn’t like a leaky roof you repair. It’s like weeds. The size of your garden determines how fast they return.

The second reason is diffusion of responsibility. On a one-person site, whoever built the button is whoever answers for it. At a big brand, the design system team built an accessible button, the product team wrapped it in something that broke it, an agency built the landing page without the design system at all, and the marketing platform injected a cookie banner over everything. Each group can truthfully say their part was fine. The user experiences the composition, and the composition is broken. Cookie consent overlays deserve special mention here: they’re often the first interactive element on the page, they come from a vendor, and when they trap focus incorrectly or can’t be dismissed by keyboard, they make everything behind them irrelevant. The brand paid for an accessible site and then put an inaccessible gate in front of it.

The third reason is that the visible layer of a big brand’s site is disproportionately made by the least accountable hands. The core product gets engineers, code review, and automated checks. The homepage hero gets an agency on a deadline, animation-heavy, rebuilt for every campaign, judged entirely on how it looks in the launch meeting. It is precisely the flashiest, most trafficked surface that routes around whatever discipline the company has. This is why you can find a rigorous, accessible account dashboard sitting behind a homepage that a keyboard can’t get past.

And there’s a fourth reason, the uncomfortable one: some large companies bought a shortcut. Overlay widgets, the products that promise to make a site accessible by injecting a script, are attractive to exactly the kind of organization that wants a line item instead of a practice. Whatever one thinks of any particular product, a script that runs after the page loads cannot restructure bad markup into good semantics, and many users of assistive technology report that such widgets interfere with the tools they already have configured. An organization that believes the problem is handled stops looking at it, which is how the problem becomes permanent.

For a small company, I think the lesson lands differently than you’d expect. The comforting read is that if the giants can’t manage it, no one can. The correct read is the opposite: the giants fail for structural reasons, too many hands, too much churn, too much distance between builder and consequence, and a small company has none of those problems. You have one site, few pages, and the person reading this probably has commit access. The famous failures are not evidence that accessibility is hard. They’re evidence that it doesn’t survive being nobody’s job.

That’s also why I built continuous scanning into GazeSite rather than one-time reports: the pattern in the big-brand data is that sites don’t stay fixed, they get re-broken by the next deploy. But tooling aside, the principle is simple enough to state in one line. Accessibility isn’t something a website has. It’s something a team does, every time they ship. The brands that get this right aren’t the ones with the biggest budgets. They’re the ones where someone specific would be embarrassed.