Percelia BFSG study 2026
How accessible are Germany's online shops?
We ran an automated scan of the homepages of 55 well-known German online shops, using the same Layer-1 analysis as our free scan (axe-core + contrast measurement). The result: 95% had at least one automatically detectable barrier. This is a quick test, not a full audit, and we set out the methodology openly.
Published:
Methodology
The study is deliberately simple and reproducible. It measures only what can be established automatically and without interaction:
- Sample: well-known German B2C online shops and consumer services across the retail categories the BFSG covers. A selection of recognisable brands, not a random draw from a register.
- Only the public homepage was tested, loaded read-only. No login, no forms, no interaction, no order.
- Analysis: Layer 1 (axe-core plus our own contrast measurement) on the rendered DOM, that is after JavaScript has run, as assistive technology sees the page.
- No AI judgement, no manual review. That is exactly the difference from a full audit.
- A shop counts as “with a barrier” if Layer 1 reports at least one violation. Shops that would not load (bot protection, timeout) are excluded from the denominator.
The most common barriers by WCAG criterion
Across all shops, the findings concentrate on a few success criteria. Each links to our reference page with an explanation and how we test it.
| WCAG success criterion | Level | Shops affected |
|---|---|---|
| 1.4.3 Contrast (Minimum) | AA | 38 / 55 (69%) |
| 4.1.2 Name, Role, Value | A | 34 / 55 (62%) |
| 2.4.4 Link Purpose (In Context) | A | 13 / 55 (24%) |
| 3.1.1 Language of Page | A | 11 / 55 (20%) |
| 2.5.8 Target Size (Minimum) | AA | 11 / 55 (20%) |
| 1.1.1 Non-text Content | A | 10 / 55 (18%) |
| 1.4.1 Use of Color | A | 9 / 55 (16%) |
| 1.2.2 Captions (Prerecorded) | A | 8 / 55 (15%) |
Distribution by severity
axe-core rates every reported violation. The large majority of findings sit in the “serious” band, that is real obstacles to operating the page.
- Critical153 (7%)
- Serious2176 (93%)
- Moderate4 (0%)
- Minor0 (0%)
By sector
For orientation, the average number of Layer-1 findings per homepage by category (only categories with at least three shops tested; small samples, so indicative only).
| Sector | Tested | with a barrier | Avg barriers |
|---|---|---|---|
| Drugstore & beauty | 3 | 3 | 87.3 |
| Department store / marketplace | 4 | 4 | 60.3 |
| Sports & outdoor | 3 | 3 | 39.7 |
| Home & furniture | 8 | 7 | 38 |
| Banking | 3 | 3 | 31.3 |
| Grocery | 7 | 7 | 30.4 |
| Travel & transport | 3 | 3 | 25.3 |
| Fashion | 8 | 7 | 14.6 |
| DIY & garden | 4 | 3 | 14.5 |
| Electronics | 5 | 5 | 2.4 |
What this means (and what it does not)
That almost every homepage tested had at least one automatically detectable barrier does not mean each of these shops is acting unlawfully. A Layer-1 finding is an indicator, not a verdict. Whether a provider is covered, any exemptions and the concrete assessment depend on the individual case.
Conversely, the figure is a floor rather than an exaggeration. Automated checks only detect part of the WCAG requirements; much of it, such as meaningful alt text, keyboard operability or the clarity of error messages, can only be judged with further analysis layers and human review. The real work needed is therefore usually greater, not smaller.
So the message is not “everyone is guilty”, but: accessibility is, for most German shop homepages, not yet achieved, and even the simplest, free check makes that visible.
Limitations of the study
- Homepage only: category, product and checkout pages, where the most critical barriers typically sit, were not tested.
- Layer 1 only: no AI judgement, no keyboard, reflow, accessibility-tree or behavioural checks, no manual review.
- Not a random sample: recognisable brands, not a representative draw. The percentages hold for this sample, not automatically for “all German shops”.
- Non-loadable shops: a noticeable share of the largest shops could not be loaded automatically due to bot protection and are missing from the analysis.
- Snapshot: measured on one date; websites change continuously.
Test your own homepage
The same Layer-1 scan this study uses is available for your website for free: instant, no signup, with a WCAG and EN 301 549 mapping.
Start a free scanFrequently asked questions about the study
- How many shops were tested?
- A list of well-known German online shops and consumer services was selected. Some could not be loaded automatically due to bot protection or timeouts; the percentages refer to the homepages successfully tested.
- Does “with a barrier” automatically mean “not BFSG-compliant”?
- No. A Layer-1 finding indicates work is needed, it is not a legal verdict. Whether and to what extent a shop is obliged depends on the individual case. Conversely, the automated check only covers part of the requirements, so the real need is usually greater.
- Were individual shops assessed by name?
- No. The study reports aggregate figures, not a ranking of individual providers. The aim is the overall picture, not a pillory.
- Can I reproduce the figures?
- Yes. The methodology is described openly, the check uses the same Layer-1 scan as our public free offering, and the metrics are generated directly from the run's raw data.
Data basis: our own Layer-1 scan (axe-core + contrast) of the shop homepages. The published metrics are reproducibly derived from the scan run's raw data.