Key Findings at a Glance
Between February and July 2026, AdaScanPro ran automated accessibility scans on 5,410 small-business websites in the United States. The scans detected 42,581 WCAG violations in total. The headline numbers:
- 96% of the 5,410 sites we scanned had at least one WCAG violation.
- 47% had at least one critical violation — the severity class that can completely block a task for someone using assistive technology.
- 71% earned a grade of D or F on our 100-point scale. More than half of all sites — 54% — received an F.
- The average site had 7.9 violations on the page scanned.
- The average accessibility score was 37.8 out of 100.
Put plainly: on the small-business web, an accessible website is the exception, not the rule. Fewer than 1 site in 12 earned an A.
This article publishes the full dataset behind those numbers — the grade distribution, the ten violations we found most often (and how to fix each one), three anonymized real-world examples, and the 2026 litigation context that turns these findings from a quality problem into a financial one.
Methodology: What We Scanned and How
We want this study to be citable, so here is exactly how the data was produced — including its limits.
The scanner. Every scan used axe-core, the open-source accessibility engine that also powers Lighthouse and the accessibility audits in browser developer tools. It is the closest thing the industry has to a standard for automated WCAG testing. AdaScanPro maps axe-core results against WCAG 2.2 Level AA success criteria (a superset of the WCAG 2.1 AA standard referenced by US courts and federal rules).
The scope. Each result reflects a single-page scan, in almost all cases the site's homepage. Homepages are typically among the best-maintained pages a business has — interior pages, checkout flows, and PDF documents usually fare worse, not better.
The sample. All 5,410 sites belong to US small businesses: e-commerce stores, restaurants, local services, fitness and wellness studios, professional practices, and retailers. The sample comes primarily from our own outreach and from businesses that ran our free scanner — it is not a random sample of the web, and we make no claim that it represents all websites (see Limitations).
The period. February through July 2026. Six months, 5,410 completed scans.
The scoring. Each site receives a 0–100 score weighted by the number and severity of violations, then a letter grade from A to F. "Critical" follows axe-core's own impact classification: violations that can make content or functionality completely unusable with assistive technology.
Overall Results: Grades, Scores, and Violations
Here is how all 5,410 sites were distributed across grades:
| Grade | Sites | Share of sample |
|-------|-------|-----------------|
| A | 408 | 7.5% |
| B | 516 | 9.5% |
| C | 647 | 12.0% |
| D | 916 | 16.9% |
| F | 2,923 | 54.0% |
Three things stand out in this distribution.
First, failure is the default state. An F was not the tail of the curve — it was the single most common outcome, by a wide margin. More sites received an F (2,923) than received an A, B, and C combined (1,571).
Second, the problem is not one bad page element. At an average of 7.9 violations per page and 42,581 violations across the dataset, most failing sites fail in several independent ways at once — navigation, images, forms, and structure together.
Third, critical failures are mainstream, not rare. Nearly half the sample (47%) had at least one violation severe enough to fully block a task for a screen reader or keyboard user. That is not a cosmetic gap. That is a customer who cannot buy, book, or contact.
The 10 Most Common Violations, Explained
To understand *what* is failing, we analyzed a detailed sample of 300 scans and counted individual violation occurrences by axe-core rule. The ten most frequent:
| Rank | Violation (axe-core rule) | Occurrences in sample | What it breaks |
|------|---------------------------|----------------------|----------------|
| 1 | region | 474 | Page structure for screen reader navigation |
| 2 | link-name | 402 | Links that announce nothing |
| 3 | heading-order | 220 | The document outline |
| 4 | landmark-unique | 158 | Telling identical landmarks apart |
| 5 | image-alt | 124 | Images invisible to screen readers |
| 6 | button-name | 108 | Buttons that announce nothing |
| 7 | image-redundant-alt | 69 | Duplicate announcements |
| 8 | nested-interactive | 60 | Unpredictable interactive controls |
| 9 | label | 49 | Form fields with no name |
| 10 | empty-heading | 46 | Headings with no content |
What each one actually means for a real visitor — and how to fix it:
1. region — content outside landmarks (474 occurrences). Screen reader users do not scroll a page visually; they jump between landmarks — header, navigation, main content, footer. When content sits outside any landmark, it becomes hard to find and easy to miss entirely. *Fix:* wrap page sections in semantic elements (header, nav, main, footer) or assign the equivalent ARIA roles. This is usually a one-time template fix that resolves dozens of instances at once.
2. link-name — links without an accessible name (402). Icon-only links, empty anchors, and image links with no alt text are announced simply as "link" — no destination, no purpose. A visitor hears "link, link, link" and has to guess. *Fix:* give every link discernible text, an aria-label, or alt text on the image inside it.
3. heading-order — broken heading hierarchy (220). Skipping levels (an h1 followed directly by an h4) scrambles the outline that screen reader users rely on to skim a page, the same way sighted users skim visually. *Fix:* use heading levels sequentially and control size with CSS, not by picking a smaller heading tag.
4. landmark-unique — indistinguishable landmarks (158). Two nav regions with no labels are announced identically; the user cannot tell main navigation from footer links. *Fix:* add an aria-label to each repeated landmark type ("Main navigation", "Footer navigation").
5. image-alt — images without alt text (124). The oldest and best-known violation, still in the top five. Every product photo, banner, and infographic without alt text simply does not exist for a blind visitor. *Fix:* write descriptive alt text for meaningful images; give decorative images an explicitly empty alt attribute.
6. button-name — buttons without an accessible name (108). The hamburger menu, the search magnifier, the cart icon: if the button is icon-only with no label, assistive technology announces "button" and nothing else. These are frequently the most important controls on the page. *Fix:* add an aria-label or visually hidden text to every icon button.
7. image-redundant-alt — alt text duplicating adjacent text (69). When an image's alt repeats the caption or link text beside it, screen readers announce everything twice. It is the mirror image of missing alt: too much, in the wrong place. *Fix:* if the text is already on the page, mark the image as decorative.
8. nested-interactive — controls inside controls (60). A button inside a link, a link inside a button — nested interactive elements behave unpredictably for keyboard and assistive-technology users: focus lands in the wrong place, or one of the two actions becomes unreachable. *Fix:* restructure so each interactive element stands alone.
9. label — form fields without labels (49). Newsletter signups, search boxes, and contact forms whose inputs have placeholder text but no programmatic label. The screen reader announces "edit text" — of what, the user cannot know. This one directly interrupts revenue: an unlabeled form is an unfinishable form. *Fix:* associate a label element with every input, or use aria-label where a visible label is genuinely impossible.
10. empty-heading — headings with no content (46). Empty h2 or h3 tags left behind by page builders create phantom entries in the document outline — stops on the navigation map that lead nowhere. *Fix:* delete them, or put real text in them.
A pattern worth noticing: almost none of these are hard to fix. The top ten are dominated by missing labels, missing structure, and template-level HTML issues — the kind of mechanical corrections a developer can batch in days, not months. The gap between a D site and a B site is usually not budget. It is that nobody ever looked. Our WCAG 2.1 AA checklist walks through the full remediation sequence.
Three Real Scans, Anonymized
Aggregates can feel abstract, so here are three real results from the dataset, anonymized by sector. These are actual scan outputs, not composites.
A jewelry e-commerce store — grade F, 22 violations, 7 critical. The critical issues clustered in image-alt, label, and button-name. In practice: a screen reader user browsing this store cannot perceive the products (no alt text), cannot complete the forms (no labels), and cannot reliably operate the buttons (no accessible names). The store is, functionally, closed to that customer — while sitting in the demographic that 79% of 2026 accessibility lawsuits target.
A pet supplies retailer — grade F, 19 violations, 6 critical. Critical issues included aria-required-children (broken ARIA widget structure), image-alt, and link-name. Product images with no descriptions and links that announce nothing turn browsing into guesswork.
A pilates studio — grade F, 18 violations, 5 critical. Critical issues in select-name, label, and heading-order — concentrated, notably, around the class booking flow. A prospective client using assistive technology likely cannot book a class unassisted. For a local service business, that is the entire website's job, failed.
None of these businesses is negligent in any unusual way. That is the point: they are typical. Their grades sit exactly where the majority of our 5,410-site sample sits.
Why These Numbers Matter in 2026
A 96% failure rate would matter in any year. It matters more in 2026, because website accessibility lawsuits against small businesses are at an all-time high.
The independent litigation data, from the two firms that track it most closely:
- In 2025, plaintiffs filed 3,117 ADA website accessibility lawsuits in federal court — up 27% over 2024 (Seyfarth Shaw's annual Title III report). Counting state courts, UsableNet tracked more than 5,000 digital accessibility suits.
- UsableNet's 2026 midyear report projects roughly 6,176 cases in 2026 (+20%), with New York (1,269), Illinois (508), Florida (393), and California (390) leading in the first half.
- 79% of 2026 targets are e-commerce businesses, and 68% of defendants generate under $50 million in annual revenue. The typical defendant is not a Fortune 500 — it is a business the size of the ones in our dataset. If you run an online store, you are in the most-sued category on the internet.
- Typical costs run from roughly $5,000 demand-letter settlements to $5,000–$30,000 lawsuit settlements, plus $30,000+ in defense fees even for defendants who prevail. Our full breakdown is in the lawsuit statistics report.
Set the two datasets side by side and the picture is uncomfortable: plaintiffs' firms use automated crawlers to find violations at scale — the same class of machine-detectable issues our scanner counts — and 96% of the small-business sites we scanned have at least one. The overlap between "what gets businesses sued" and "what most small-business websites contain" is nearly total.
One more data point for anyone tempted by a shortcut: 38.5% of businesses sued over digital accessibility already had an overlay widget installed (AudioEye, 2026). Detection is not remediation, and neither is a widget.
Limitations and Honest Caveats
Original research is only useful if its limits are stated plainly. Ours:
This is not a random sample of the web. The sites came from our outreach lists and free-scan users — US small businesses, skewing toward consumer-facing sectors. The correct reading is "96% of the 5,410 small-business sites we scanned," not "96% of all websites."
One page per site. Homepages tend to be a site's best-maintained page. Full-site scans in our experience surface more issues per site, not fewer — so per-site violation counts here are likely understatements.
Automated testing has a ceiling. axe-core detects the machine-detectable subset of WCAG failures. It cannot judge whether alt text is *accurate*, whether focus order makes *sense*, or whether a keyboard-only user can complete a *full journey*. Issues automation cannot see come on top of these numbers. In every direction, the honest interpretation is the same: these figures are a floor, not a ceiling.
A scan is a diagnosis, not a cure — or a certificate. No automated tool can make a website compliant or certify that it is; the FTC's $1 million order against overlay vendor accessiBe in 2025 was precisely about that kind of claim. What a scan does is show you, specifically and page by page, where you stand and what to fix first.
What Small Businesses Should Do With This
Three steps, in order of leverage:
1. Get your baseline. You cannot fix what you have not measured, and as this dataset shows, the odds that your site is in the failing 71% are high. A scan takes about a minute.
2. Fix the template-level issues first. The top ten list above is dominated by violations that live in your site's template — landmarks, heading structure, icon buttons, nav links. Fixing them once fixes them on every page. Most sites can move up two letter grades on mechanical corrections alone.
3. Make it recurring. Sites drift. Every new page, plugin, and product photo is a chance to reintroduce a violation. The sites in our A bracket are not lucky — they are checked.
Run the same free scan we used in this study on your own website. It takes about a minute — and you will know exactly which side of the 96% you are on.
How to Cite This Study
You are welcome to cite or reference this data. Please attribute it as "AdaScanPro scan data, February–July 2026 (5,410 US small-business websites)" and link to this page. Summary statistics for quick reference:
- 5,410 websites scanned; 42,581 WCAG violations detected (automated, axe-core, single-page scans)
- 96% of scanned sites had at least one violation; 47% had at least one critical violation
- Grade distribution: A 7.5% · B 9.5% · C 12.0% · D 16.9% · F 54.0% (71% D or F)
- Average: 7.9 violations per site; mean score 37.8/100
- Most common violations: region, link-name, heading-order, landmark-unique, image-alt
For media inquiries or a deeper cut of the data, contact us.
Tags
Share this article