Research
European Accessibility Benchmark 2026
Automated accessibility testing across a stratified sample of B2B manufacturing and industrial websites in the Netherlands, Belgium and Germany.
NorthPeak scanned the primary public page of 150 B2B manufacturing and industrial companies — 50 in the Netherlands, 50 in Belgium and 50 in Germany — using the same automated accessibility scanner for every site. 135 of 150 scans completed successfully; the remaining 15 could not be reached or rendered for technical reasons unrelated to accessibility (see Methodology). This report presents the aggregate results.
Executive result
Among the 135 pages successfully scanned, 134 — 99.3% — had at least one automatically detectable accessibility issue. The automated scan detected 695 issues in total across those pages.
This does not mean 99.3% of these companies are "non-compliant" with any legal standard. Automated testing detects only a subset of WCAG success criteria — typically estimated at around a third — and cannot itself establish WCAG or European Accessibility Act conformance. What it does show is that automatically detectable issues are common, not rare, among this sample.
Severity of detected issues
Of the 695 automatically detected issues, 109 were classified critical, 274 serious, 281 moderate and 31 minor by the underlying axe-core rule engine's own impact classification.
Manual-review flags — a separate category
The scanner additionally flagged 229 conditions across 129 pages (95.6% of successfully scanned pages) as requiring human verification — for example, a colour-contrast ratio the tool cannot fully evaluate without knowing the intended visual design, or an ARIA pattern that is technically valid but may not behave correctly for assistive technology.
These are not counted among the 695 automatically detected issues above, and are not confirmed accessibility violations. They indicate where a manual expert review would be needed to reach a conclusion — see Limitations.
Country comparison
Because every country sample shares the same predefined industry-category and company-size composition (see Methodology), the country figures below are directly comparable to one another — though not to the wider population of manufacturers in each country, since this is a stratified sample, not a random one.
Detected-issue prevalence was similar across all three countries: Netherlands 100% (44 of 44 successfully scanned); Belgium 97.8% (44 of 45 successfully scanned); Germany 100% (46 of 46 successfully scanned).
Company size and industry category
Detected-issue prevalence was consistently high across every company-size band and every industry category in this sample (see the tables below) — no size band or category stood out as meaningfully better or worse than the others. With prevalence already near-universal in every group, this sample is not large enough to responsibly draw further conclusions about which specific company profiles are more or less likely to have automatically detectable issues.
Automated detection is not legal compliance
No automatically detected issue count, and no absence of one, establishes or rules out WCAG or European Accessibility Act conformance for any page. See Limitations below.
Check your own site
Curious how your own company's website measures up — not just on accessibility, but on export readiness, technical product presentation and B2B conversion too? Use the free B2B Manufacturing Website Readiness Scorecard 2026.
Use the free Scorecard| Country | Successfully scanned | Pages with ≥1 detected issue | Prevalence |
|---|---|---|---|
| Netherlands | 44 | 44 | 100% |
| Belgium | 45 | 44 | 97.8% |
| Germany | 46 | 46 | 100% |
| Issue | Affected pages | % of successfully scanned pages |
|---|---|---|
| Insufficient colour contrast | 103 | 76.3% |
| Content not contained in a landmark region | 93 | 68.9% |
| Links without discernible text | 90 | 66.7% |
| Heading levels skip a level | 56 | 41.5% |
| Duplicate landmark roles/labels | 50 | 37% |
| Buttons without discernible text | 31 | 23% |
| Images missing alternative text | 27 | 20% |
| No single main landmark | 25 | 18.5% |
| Zooming/scaling disabled | 23 | 17% |
| Unsupported ARIA attributes | 16 | 11.9% |
| Size | Successfully scanned | Pages with ≥1 detected issue | Prevalence |
|---|---|---|---|
| Small (< 50 employees) | 26 | 26 | 100% |
| Medium (50–249 employees) | 68 | 68 | 100% |
| Large (250+ employees) | 41 | 40 | 97.6% |
| Category | Successfully scanned | Pages with ≥1 detected issue | Prevalence |
|---|---|---|---|
| Machinery & industrial equipment | 28 | 28 | 100% |
| Metal / components / fabrication | 26 | 26 | 100% |
| Electrical / electronics / automation | 26 | 26 | 100% |
| Materials / chemicals / industrial supplies | 27 | 26 | 96.3% |
| Other manufacturing / technical B2B | 28 | 28 | 100% |
Methodology
Sample and selection
150 companies were selected — 50 per country in the Netherlands, Belgium and Germany — using a deterministic, stratified sampling procedure fixed before any scanning began. Companies were drawn from B2B manufacturing and industrial sectors (machinery, metal/fabrication, electronics/automation, materials/chemicals, and other technical B2B manufacturing), with each country sample built to the same category and company-size quota so the three country samples are directly comparable to one another.
This is a predefined, purposively stratified benchmark sample — not a random or population-representative sample of all manufacturers in each country, and it should not be described as either.
Scanning
Each company's selected public page — normally its primary homepage — was tested once, using the same automated accessibility scanner, the same axe-core rule engine (version 4.13.0) and the same browser engine (Playwright 1.62.1) for every site. Scans ran sequentially, one at a time, at a conservative pace — never as a bulk or high-concurrency crawl. A pre-defined single retry was permitted only for a transient technical failure (a timeout or temporary connection error) — never because of the findings a scan returned.
What counts as a technical failure
15 of 150 scan attempts (10%) did not complete — for example, a site blocking automated access, a TLS certificate error, or a page that did not finish loading within the scanner's timeout. These are excluded from every prevalence percentage in this report, which is calculated only among the 135 successfully scanned pages — never against all 150 selected companies.
Scope of a single scan
Each result represents one automated scan of one selected page — not a crawl of the whole website, and not a manual audit. Automated tools reliably detect only a portion of WCAG success criteria; many requirements (meaningful reading order, correct focus management, accurate alternative text, keyboard-operability of custom components, and more) require manual expert review and testing with assistive technology, and are outside what this benchmark measures.
More from NorthPeak
Want to see what automated testing finds on your website?
Run a free accessibility scan of one public page. For a complete WCAG assessment, manual expert review is still required.