# PDF Accessibility Audit -- Aggregate Findings from a Real Survey (August 2026)

This is not a mockup and not a single client's report. It's the real, aggregate output of a survey Reologix ran in August 2026 against 80 seeded public-agency websites, split across four separate populations: suburban park districts, downstate/rural park districts, public libraries, and municipalities. Every number below comes from an actual structural scan of a real, publicly posted PDF -- nothing is estimated or rounded up.

## The Sample, Honestly

- **Entities seeded across all four populations:** 80 (20 per population)
- **Entities that actually yielded documents we could audit:** 41. The rest were blocked by the site's own defenses, unreachable, or where our crawl didn't reach a public document library -- we only count what we actually scanned.
- **Total files audited:** 1,349
- **Total pages audited:** 36,587

## By Population

Rates differ enough between populations that a single blended number would hide the real picture, so we report each one separately.

| Population | Entities seeded | Entities with data | Files | Pages | Fails at least one check |
|---|---|---|---|---|---|
| Park districts (suburban) | 20 | 12 | 400 | 11,446 | 91.8% |
| Park districts (rural / downstate) | 20 | 13 | 407 | 6,006 | 96.3% |
| Public libraries | 20 | 9 | 348 | 4,519 | 83.3% |
| Municipalities | 20 | 7 | 194 | 14,616 | 99.0% |

## A Tag Tree Is Not the Same as a Passing Document

"Untagged" is the most severe failure, but it isn't the only one, and fixing it doesn't mean a document is done. Across every entity in this dataset where every single file already had a tag tree in place, 95% of pages still carried at least one accessibility failure -- almost always a missing title or missing image descriptions. If your documents "came out of Word" and you've assumed that settles it, this is the number that says otherwise.

## All 10 Failure Types Found, Ranked by How Often They Appear

Across all 1,349 files audited:

| Finding | Files affected | Share of all files |
|---|---|---|
| No document title set | 539 | 40.0% |
| Images missing text descriptions (alt text) | 485 | 36.0% |
| No tagging structure at all (untagged document) | 462 | 34.2% |
| Not set to display the document title in the viewer | 434 | 32.2% |
| Links give no destination information ("click here") | 261 | 19.3% |
| Tables missing header labels | 199 | 14.8% |
| No document language set | 181 | 13.4% |
| Not marked as a tagged document | 173 | 12.8% |
| Heading levels skip a step | 63 | 4.7% |
| No selectable text at all (scanned image, needs OCR) | 5 | 0.4% |

## What This Means

In every one of the four populations above, the share of files failing at least one check ranged from 83% to 99% -- far more than the "completely untagged" figure alone would suggest in any of them. Untagged and scanned-image files are the highest priority, since assistive technology can extract nothing meaningful from them at all. Files with only a few issues (a missing title, a handful of images without descriptions) are generally quicker fixes -- but as the table above shows, very few files have *zero* issues.

We don't publish which specific agencies these numbers came from -- these are prospects, not case studies. A detailed, file-by-file breakdown of your own documents is exactly what you get in your own audit's `results.csv`.