Blog
April 2027: which of your PDFs the ADA rule actually covers
The Title II deadline moved to 26 April 2027. Not every PDF on your site is in scope, and the exception turns on use rather than age.
State and local government entities serving 50,000 people or more have until 26 April 2027 to meet WCAG 2.1 Level AA on their web content, and that includes the PDFs they publish. Smaller entities and special districts have until 26 April 2028. The part most summaries skip is that not every PDF on your site is in scope: older documents are excepted, but the exception turns on whether people still use the document, not on how old it is. So the first piece of work is a triage, not a remediation.
The dates, including the one that moved
The Department of Justice published its Title II web accessibility rule in the Federal Register on 24 April 2024. The original compliance dates were April 2026 and April 2027. On 20 April 2026 the Department issued an interim final rule extending both by a year.
| Who | Compliance date |
|---|---|
| State and local government entities with a total population of 50,000 or more | 26 April 2027 |
| Entities with a total population under 50,000, and any special district government | 26 April 2028 |
The technical standard in both cases is WCAG 2.1, Level AA.
“Public entity” is broader than people expect. It covers counties, cities, towns, public school districts, public colleges and universities, public libraries, transit authorities, water districts and other special districts. If you are a private organisation this rule is not yours, though Title III litigation and, in the European Union, the European Accessibility Act are separate pressures in the same direction.
What the rule asks of a PDF
WCAG was written for web pages, and applying it to a PDF means the document has to carry structure a screen reader can use. In practice:
- Tags. A parallel tree inside the file that says this run of text is a level two heading, this block is a list, this is a table with these rows and columns. Visual formatting is not structure. Text that merely looks like a heading is not one.
- Reading order. The order the tags run in, which is what a screen reader follows, and which can differ completely from the visual layout on a multi column page.
- Alternative text on images that carry meaning.
- Real table structure, with header cells identified as header cells.
- A declared document language, so the right voice is used.
- Meaning not carried by colour alone, and sufficient contrast.
There is a floor below all of that. A PDF that is a scan, with no text layer at all, cannot meet any of it. There are no characters in the file to tag. Those documents are not a formatting problem, they are an OCR problem first and a tagging problem second, and they are the most expensive things in any back catalogue. Our post on telling a real text PDF from a photo of one covers how to spot them.
The exceptions, and the one that matters
The rule has five exceptions. Read them properly, because they are the difference between remediating four thousand documents and remediating four hundred.
| Exception | What it covers |
|---|---|
| Archived web content | Content that is pre-compliance-date or reproduces physical media, and kept only for reference, research or recordkeeping, and stored in a designated archive area, and unchanged since it was archived. All four. |
| Preexisting conventional electronic documents | Word processing, presentation, PDF and spreadsheet files that were already on the site before your compliance date. |
| Third party content | Material posted by members of the public that you do not control, such as message board posts. |
| Individualised password-protected documents | A document about one person, property or account, in one of those file formats, behind a login. A specific person’s water bill, not the standard form. |
| Preexisting social media posts | Posts made before your compliance date. |
The second row is the one that decides the size of the job, and it has a carve out that is easy to miss. A preexisting document loses the exception if it is currently being used to apply for, gain access to, or participate in your services, programs or activities.
The Department’s own example: a state posted a PDF business licence application in 2020, and people still use that PDF to apply for a licence after the compliance date. The exception does not apply. The document has to meet WCAG 2.1 AA.
So the test is not “when did we publish this”. It is “is anyone using this to get something from us”. A 2019 council meeting agenda that nobody opens is excepted. A 2019 benefits application form that is still the way you apply for benefits is not.
Note too that the archive exception is not a state of mind. It requires a designated archive area and that the content has not changed since it went there. Calling a page “archive” in a breadcrumb does not do it, and if somebody has edited a document since it was archived, it is not archived any more.
The triage
The actual first piece of work is counting, and almost nobody has the number. A medium sized city typically publishes somewhere in the thousands of PDFs and has no list of them.
- Crawl the domain and list every PDF. Every linked one and, if you can, the ones sitting in the uploads directory that nothing links to any more. You will find more than you expected.
- Sort by use, not by date. Which of these are forms, applications, permits, registration documents, notices people act on, or anything the public needs in order to get something? That set is in scope regardless of age. Everything that is genuinely reference-only and properly archived is a candidate for the exception.
- Within the in-scope set, check for a text layer. A file with no text at all needs OCR before anything else and costs several times more to remediate. You want that number separately, because it is the one that breaks budgets. Our converter reports this per file, in your browser, without uploading anything.
- Then check tagging. Of the files that do have text, how many are tagged at all, and how many are tagged usefully. This needs an accessibility checker rather than a converter.
The output of that exercise is a sentence like “we publish 4,231 PDFs, 1,180 are in scope, 260 of those are image-only scans.” That sentence is what a budget request is made of, and it is worth producing before April rather than after.
Two things worth doing now regardless
Stop making new ones. Every PDF published from now until the deadline is a document published after your compliance date, so no exception applies to it at all. Authoring accessibly in Word or InDesign and exporting with tags costs almost nothing compared to remediating later. Fixing the pipeline first stops the queue growing while you work on it.
Consider whether it needs to be a PDF. The most efficient remediation for a form is often an HTML form. The rule is about the content being accessible, not about preserving the format it happens to be in, and an accessible web page is usually cheaper to build and maintain than an accessible PDF.
One caution on merging: combining PDFs strips the tag tree, so a merged bundle is untagged even if every input was tagged. If you assemble documents before publishing them, merge first and tag afterwards, never the other way round.
Further reading
- Fact sheet: new rule on the accessibility of web content and mobile apps provided by state and local governments, ADA.gov. The exceptions are set out in the Department’s own words.
- First steps toward complying with the Title II web rule, ADA.gov.
- WCAG overview, W3C Web Accessibility Initiative.
- Section508.gov guidance on creating accessible PDFs, which is written for federal agencies but is the clearest practical checklist available.
- The European Accessibility Act, for organisations with obligations in the European Union as well.