The sample
- 63 organisations, 180 PDFs, 4740 pages.
- Sectors: city, county, school district, community college, university, state agency.
- Results are shown in aggregate. No organisation is named.
What we found
- 13.3% of PDFs had no tags, so a screen reader cannot read them in order.
- 33.9% had no document title.
- 17.8% had no document language set.
- 31.1% had images with no alternative text.
- 28.7% of the tagged documents of two or more pages (122 documents) had no heading structure.
- 33.8% of the documents that contain tables (68 documents) had tables with no header cells.
- 3.3% (6 PDFs) contained scanned, image-only pages.
By sector
| Failure | All | City | County | School district | Community college | University | State agency |
|---|---|---|---|---|---|---|---|
| No tags, so a screen reader cannot read the document in order | 13% | 17% | 39% | 20% | 8% | 0% | 11% |
| No document title | 34% | 39% | 22% | 65% | 28% | 30% | 31% |
| No document language set | 18% | 17% | 44% | 20% | 11% | 4% | 18% |
| Images with no alternative text | 31% | 39% | 28% | 30% | 44% | 17% | 28% |
| No heading structure (tagged documents of 2 or more pages) | 29% | 43% | 46% | 27% | 29% | 21% | 24% |
| Tables without header cells (documents that contain tables) | 34% | 50% | 40% | 62% | 21% | 56% | 21% |
| Scanned, image-only pages | 3% | 0% | 11% | 5% | 0% | 4% | 3% |
| PDFs in the sample | 180 | 18 | 18 | 20 | 36 | 23 | 65 |
| Organisations | 63 | 6 | 6 | 7 | 12 | 9 | 23 |
Small groups (a few documents) move a lot with one file. Read sector columns with the PDF counts in the last rows.
Methodology
- Sample: PDFs linked from public websites of the organisations, fetched in October 2026. A few PDFs per organisation.
- Checks: automated checks of machine-detectable PDF/UA and WCAG 2.1 AA criteria. A failure is counted once per PDF, however many times it occurs in that PDF.
- Denominators: percentages are of all PDFs, except the heading line (tagged documents of two or more pages) and the table line (documents that contain tables).
Limitations
- This is a small, non-random sample. It describes these files, not every public body or every document.
- Automated checks find only machine-detectable failures. Reading order, colour contrast, meaningful alternative text and many other criteria need human review, so a file with no failure here may still be inaccessible.
- A result is not a conformance determination, not a compliance finding and not legal advice.
Get your organisation’s own results
If your organisation is in the sample, or you would like your own public PDFs checked, request the results. We send you the findings for your files only, by email.
Request your organisation’s results
Opens your email app, addressed to contract@trilatic.com.
Buying remediation? Use the free PDF acceptance criteria for public buyers to test the work. Want the first findings fixed? See accessible PDF remediation or the ADA Title II document explainer.
League tables by state
See how public bodies in each state score on the same kind of checks: PDF accessibility league tables. Or grade one PDF yourself with the PDF Report Card.