What a website scan reveals about the PDF collection of a mid-sized town: an average of 19 out of 100 points, missing structure in 92 documents – and how such findings turn into a workable order of priorities.
Anonymised case study from a real processing run; the name of the organisation and the document title have been changed.
In the Pre-Check of a mid-sized town, 100 PDF documents from the website scored an average of 19 out of 100 points; 89 of them were critical. That sounds dramatic, but it is a typical picture – and largely solvable automatically: 90 of the 100 documents could be improved automatically, and after optimisation the selected documents scored an average of 100 points. The real work lay elsewhere: in deciding where to start, and in actually getting the improved files onto the website.
Starting point
The starting point was the question many public bodies ask themselves: what state are the PDFs on our website in? That is what the Pre-Check is for: a portal, accessible via a personal link, that shows and improves the PDF collection of a website in five steps.
- Website scan: a crawler follows the links of the website and records every linked PDF file together with the page on which it was found.
- Check: every PDF is checked and receives a score as well as a rating of critical, medium or good.
- Selection: the administration sees the results, ordered by the structure of the website if desired, and selects what is to be improved.
- Optimise: the selected documents are remediated automatically, with AI only when needed. The originals remain unchanged.
- Adopt: the improved versions are available as a ZIP file for download and replacement in the content management system – or, where a connection exists, as a direct replacement on the website.
What the check found
Overall result of the Pre-Check| Metric | Value |
|---|
| Documents checked | 100 PDFs with a total of 1,951 pages |
|---|
| Average score | 19 out of 100 |
|---|
| Rating | 89 critical, 3 medium, 7 good |
|---|
| Automatically improvable | 90 documents |
|---|
The most common findings, counted by documents affected:
Most common findings in the Pre-Check| Finding | Number | What it means for users |
|---|
| Structure (tags) missing or incomplete | 92 | Headings, lists and tables cannot be recognised by screen readers. |
|---|
| Document title missing | 89 | The file name is displayed and read aloud. |
|---|
| Document language missing | 71 | The screen reader has to guess the pronunciation. |
|---|
| No bookmarks | 40 | Longer documents lack navigation. |
|---|
| Fonts not embedded | 26 | Display and text output depend on the computer. |
|---|
| Scanned pages without text | 16 | For screen readers, the page is empty. |
|---|
| Tab order missing | 5 | Links and fields are reached by keyboard in the wrong order. |
|---|
| Images without alt text | 4 | Only ‘graphic’ is announced. |
|---|
| Headings not marked up | 4 | No navigation via headings. |
|---|
Important when reading this: the low numbers for alt text and headings do not mean that everything was in order there. Where the structure is missing entirely, there are also no figure or heading tags to which a checking tool could attach these errors. Such problems only show up once the structure has been built.
What DokAudit did automatically
For the selected documents, the optimisation carried out the safe repairs that each document needed: building the structure, for scanned pages after prior text recognition; setting title and language; generating bookmarks from the headings; embedding fonts; aligning the tab order with the structure; adopting captions as alt text and generating suggestions for the remaining images. DokAudit then checked every document again with veraPDF and its own checks of the logical structure. Afterwards, the selected documents scored an average of 100 points.
What a person still had to do
- Select and prioritise: first forms, applications, by-laws and fee schedules in force – they are needed for administrative procedures. Then the current official gazette (Amtsblatt) and notices of ongoing procedures. Then whatever is downloaded most often according to the web statistics. Documents that only lack title and language are quick wins.
- Caution with the archive exemption: many files are older than 23 September 2018. They are only exempt if they are not needed for active administrative processes; an old by-law that is still in force is therefore generally not covered. Declaring the entire legacy collection an archive across the board will not stand up to scrutiny. How to distinguish properly is explained in Keeping your website's PDFs under control for good.
- Review suggestions: approve or improve AI suggestions for alt text, and listen to spot checks of the reading order of multi-column documents.
- Replace: replace the improved files at the same address wherever possible, so that links and search results remain valid.
- Update the statement: update the accessibility statement with the new status and the documents deliberately deferred.
Result
An average of 19 points became an average of 100 for the selected documents. The Pre-Check also provided a reliable basis for planning: the town knows which files exist, where they are linked, what they lack and what can be fixed automatically.
Lessons learned
- A low average is normal. It arises mainly from missing basics – structure, title, language – and those are exactly what can be added automatically.
- Missing structure hides other problems. Only once the structure has been built do alt text and headings really come to the table.
- Prioritise first, then optimise. Relevance to procedures and downloads are better criteria than the age of a file.
- Replacement is the step that is most often left undone. A repaired file that never reaches the website helps nobody.
- Once is not enough. New PDFs are added every week. Without ongoing monitoring, you will be back at square one in a year – that is what Continuous Accessibility, the regular automatic check of the website, is for.
You can start the free website scan for your own website.
Sources:
As of: 10/2026. This article gives a general overview and is not legal advice. The legal and standards texts in force are authoritative; in individual cases, the law of the German states (Länder) may differ.