Before and after: a county's annual financial report, remediated and verified
An annual comprehensive financial report (ACFR) is the hardest document a local government publishes to make accessible: hundreds of pages, dense financial tables, and an audit trail that can't change. Here is one from a Florida county, before and after PairedUp remediated it, with every test we ran and every number it produced.
By Wesley Black, founder of PairedUp · Published October 22, 2026
A 305-page report with no structure at all.
The county's FY2025 ACFR, as published, had no tags. To a screen reader it was one long stream of text and numbers with no headings, no tables, and no reading order. That is typical. In our study of 481 Florida government documents, all 249 ACFRs failed automated accessibility testing.
failures in veraPDF, almost all from content that was never tagged and text with no declared language.
failures. This is the checker most offices use, and it makes a broken file look nearly fine.
editions we tested (FY2023 to FY2025) failed the same way. Nothing carried forward.
Zero failures, and the table still says what it said.
We remediated a three-page excerpt chosen to cover each kind of content in the report: the formal title page, a page of management's discussion and analysis, and the General Fund schedule of revenues, expenditures, and changes in fund balance, budget and actual. That schedule has about 60 department rows across four amount columns (original budget, final budget, actual, and variance). Financial tables like this are where most remediation goes wrong, so we check every cell.
| Check | What it tests | Result after remediation |
|---|---|---|
| PDF/UA machine checks | veraPDF 1.30.2, PDF/UA-1 profile: every machine-verifiable requirement of the PDF accessibility standard | 0 failures, 27,012 checks passed |
| Acrobat-rule checks | PairedUp's implementation of Adobe Acrobat's published accessibility rules (not Adobe software) | 0 failures: 20 passed, 2 flagged for manual review, 10 not applicable |
| Table content check | Every numeric cell is read back out of the tagged file and compared with the printed value and its row and column headers | 296 of 296 cells correct |
| Visual check | Each page is rendered before and after and compared pixel by pixel, so the audited document looks exactly the same | 0 pixels changed on all 3 pages |
| Structure check | The tag tree contains real structure bound to real content, not a placeholder that passes validators | Headings, 1 data table with header and data cells, decorative content marked as artifacts |
The last two checks exist because a file can pass validators while being useless to a person. A tag tree of empty paragraphs will satisfy some checkers. We only count a document done when the structure is real and the content inside it is unchanged.
Passing the checks is necessary. It is not the whole job.
No automated tool can decide whether alt text describes an image well or whether a reading order makes sense to a listener. WCAG 2.1 AA conformance also depends on those judgment calls, and they're best made by someone who knows the document. We describe our results as exactly what they are: documents that pass the tests named on this page, which you can re-run yourself with free tools.
Your staff makes the judgment calls
We list the pages where image descriptions, reading order, table headers, link text, or color contrast need a human look. Our free review checklist shows how to check each one in minutes.
Why the county isn't named
We don't publish any government's failing results by name. The original is a public record, and we're glad to walk through both files on a call.
Check it yourself
veraPDF is free and open source. Run it on your own ACFR and compare the count with the 76,132 above.
See the same before and after on your own report.
Tell us your city or county. We'll build a private demonstration from your own published ACFR or budget book before we meet, at no cost, with every test result attached.
Book 30 MinutesPick a time. That's it.