Web accessibility by industry, 2026

Across 342 homepages in four sectors, ranked by the share with at least one critical issue (the kind that can hard-block a screen-reader user), lower is better:

Government came out most accessible and online stores least, with a nearly threefold gap in critical-issue rate between them (18% versus 49%). The three non-retail sectors cluster tightly on the broad "any serious failure" measure, all around two in three, but pull apart sharply on the severe end and on sheer issue volume, where the median store had more than twice the issues of the median government site. Full write-ups: government, banks and finance, hospitals, online stores.

Before the differences, the thing they share. Across all 342 sites, 71% had at least one serious failure on the homepage, and only 10% passed every automated check. Even the best-performing sector still shipped a serious failure on two-thirds of its homepages. So this is a ranking of degrees, not a leaderboard with a winner anyone should be proud of. The gap between first and last is real and worth understanding, but first place is still a long way from good.

The interesting part is what does not explain the ranking. It is not money or legal exposure. Banks have both in abundance, more of each than a hospital or a government agency, and they landed in the middle. Online retail is the most commercially sophisticated sector of the four and came last. If budget and lawsuit risk decided this, the order would be reversed.

The factor that tracks the ranking is duller and more useful: whether a sector builds on a shared, accessible component system. Government does, the US Web Design System, and it is both the cleanest sector and the one whose pages look most standardised to a scanner. The clearest evidence is a number that has nothing to do with pass rates:

• Median items needing human review: 2 for government, 2 for banks, but 15 for hospitals and 24 for online stores.

A low "needs review" count means predictable, semantic markup that an automated test can actually reason about, the fingerprint of a design system used consistently. The sectors built from shared components (government on the USWDS, banks largely on enterprise platforms) are not just cleaner, they are more legible. The sectors that hand-roll every page (retail especially) are messier by every measure. The lesson for any team is the boring one: you get more accessibility from adopting an accessible component library than from any amount of after-the-fact auditing.

Whatever the sector, the top of the list barely changed. Across all 342 sites, the most common WCAG failures were:

• Colour contrast (30%) — WCAG 1.4.3. The single most common failure in every sector we tested, and our figure includes text on CSS gradients that most scanners skip.
• Touch target size (20%) — controls under 24×24px, WCAG 2.5.8.
• Images missing alt text (14%) and links with no discernible text (14%) — WCAG 1.1.1 and 2.4.4 / 4.1.2.
• Buttons with no accessible name (11%) — WCAG 4.1.2.

Contrast and target size were the top two in every single sector. That is the encouraging part of an otherwise grim picture: the most widespread barriers on the web are also among the most mechanical to find and fix. A design pass on colour contrast and hit-target sizes alone would move the numbers in every industry here.

None of this is a legal verdict. Automated failures are not a compliance determination, we tested homepages rather than whole sites, and we have named no individual organisation in any sector. What the combined data supports is narrower and more practical: accessibility on major US websites is still the exception rather than the rule, the sectors that do best are the ones building from shared accessible components, and the most common barriers are also the most fixable.

Those fixes belong in the code itself, not in a layer added on top of it. You can see where your own site stands with the free scan, or wire it into CI with the CLI and GitHub Action so it does not regress after the next release.

In the interest of not overclaiming:

• Sample: 342 homepages that returned a real page, out of 391 attempted, across four curated sector samples (online stores, US health systems, US federal government, and US banks and financial services), scanned across mid-2026. The excluded 49 mostly returned bot-protection blocks or protocol errors; the exclusion rate was far higher for banks and government than for stores and hospitals, which is discussed in those individual reports. These are curated samples of recognisable organisations, not random or traffic-ranked ones, so read the ranking as a comparison of well-known sites in each sector, not a census.
• Engine: the axe-core rules for WCAG 2.x Level A and AA, plus our own resolution of colour contrast over CSS gradients, applied identically to every sector so the numbers are comparable.
• Scope: homepages only. Checkout, patient portals, government application flows, and online banking sit behind the homepage and often fare worse.
• Not a compliance audit: detectable WCAG failures are not a legal determination, and automated testing covers only the machine-checkable part of WCAG, roughly a third of the success criteria. The real barrier rates are almost certainly higher than the figures here, not lower. We report aggregate figures only.