Portfolio demo. Payments run in Stripe test mode: use card 4242 4242 4242 4242, any future date and any CVC. No real charges.

ScanAble Research

The 2026 State of Web Accessibility: We scanned the world's most-visited sites

Published · Data refreshed

We ran an automated WCAG 2.0/2.1/2.2 AA audit against the top 500 most-visited websites in the world (as ranked by Tranco). Of those, 296 scanned successfully — 201 blocked us at the network layer or failed to render, which is itself one of the more interesting findings: top sites increasingly treat headless browsers as adversarial traffic, and that has implications for any team trying to monitor accessibility from a CI pipeline. Here’s what the audit-able subset says about the state of web accessibility going into the back half of 2026.

Average score

42.5

Median score

36.5

% scoring below 70

78%

Worse than expected

% with a critical issue

44%

The headline

Of the top sites that let us in, the average compliance score is 42.5 out of 100, and 44% ship at least one critical accessibility violation — the kind that blocks a screen-reader user from completing a basic action like submitting a form or following a link with no accessible name.

78% score below 70, the threshold a site needs to clear before any meaningful share of WCAG Level AA rules pass cleanly.

The same handful of issues account for the bulk of the failure. Three rules — color-contrast (40% of sites), region (38%), and landmark-one-main (28%) — show up on the majority of audits we ran.

Score distribution

Where the world’s top sites land

90-100
27 (9%)
70-89
38 (13%)
50-69
44 (15%)
30-49
79 (27%)
0-29
108 (36%)
The repeat offenders

Top 10 violations across the audit set

These are the axe-core rules that fail on the largest share of sites we scanned. The first column is the rule ID; the percentage is the share of audited sites that failed it at least once.

RuleSeverity% of sites failing
color-contrast

Ensure the contrast between foreground and background colors meets WCAG 2 AA minimum contrast ratio thresholds

serious40%
region

Ensure all page content is contained by landmarks

moderate38%
landmark-one-main

Ensure the document has a main landmark

moderate28%
link-name

Ensure links have discernible text

serious25%
image-alt

Ensure <img> elements have alternative text or a role of none or presentation

critical20%
heading-order

Ensure the order of headings is semantically correct

moderate19%
page-has-heading-one

Ensure that the page, or at least one of its frames contains a level-one heading

moderate18%
html-has-lang

Ensure every HTML document has a lang attribute

serious17%
target-size

Ensure touch targets have sufficient size and space

serious14%
landmark-unique

Ensure landmarks are unique

moderate14%

Violations by severity

Across all 296 successfully-scanned sites, the scanner counted 11,936 total violations. The breakdown:

Critical

1,279

Serious

4,878

Moderate

5,441

Minor

338

That’s an average of 40.3 violations per site.

What it means for ADA & EAA compliance

My honest read of the data: this isn’t a knowledge problem. The top 10 violations are entry-level WCAG 2.0 rules — color contrast, missing landmarks, links without text, images without alt. Anyone shipping a frontend in 2026 knows what these are. Some have known for ten years.

What I think it actually is: most engineering teams haven’t shipped accessibility into the default deploy workflow. They check it manually before launch (sometimes), they fix the obvious things in a sprint when a stakeholder asks (usually), and then it drifts. Compare that to security: you don’t ship a feature without it going through the security pipeline, because the pipeline exists and runs on every commit. Accessibility doesn’t have that pipeline at most shops, so it stays a Q4 project.

The fix isn’t more auditors or more expensive consultants. It’s a default check that runs on every deploy and produces a number anyone — engineer, PM, compliance officer — can read. That’s the thesis behind selling the same scanner as a $19/month API. The PDF is the artifact for procurement; the API is the workflow for the team that’s going to keep it clean once they care.

Methodology

How we generated these numbers

  • Source list: the top 500 domains from the Tranco ranked list of most-visited sites.
  • What we scanned: the homepage of each domain (https://<domain>/), loaded in headless Chrome via Puppeteer.
  • Engine: axe-core with WCAG 2.0, 2.1, and 2.2 Level AA tags enabled. Same engine ScanAble uses for every paid audit.
  • Score formula: passes / (passes + Σ(violation_nodes × severity_weight)) × 100. Full breakdown at /methodology.
  • Caveat: automated tools (including ours) catch ~30–40% of WCAG issues per Deque’s research. The numbers above are a floor, not a ceiling — manual testing would surface more.
  • Data file: regenerated by scripts/analyze-domain-scans.ts. The article reads app/research/data/state-of-web-accessibility.json at build time, so re-running the batch and re-deploying updates these numbers in place.

Find out where your site lands

Same scanner, same engine, same scoring. URL in, signed PDF audit out in 60 seconds.

Run a free previewSee the methodology