What to Do With 10,000 Inaccessible PDFs

Every accessibility program reaches the same uncomfortable moment. The website work is scoped, the audit is underway, and someone asks about the documents.

Then you count them. It's rarely hundreds. It's usually thousands, sometimes tens of thousands — agendas, forms, reports, notices, statements, policies, a decade of board packets. Many of them scanned from paper, which means they're images of text with no text at all.

Why one-at-a-time never finishes

Do the arithmetic before committing to a remediation queue. Meaningful remediation of a complex document — tag structure, reading order, alt text, table headers, form fields — takes a practitioner somewhere between half an hour and several hours.

At 8,000 documents, that's years of effort. Meanwhile you're publishing new ones every week. The backlog regenerates faster than it clears, and you're funding a permanent remediation line item that never reaches the end.

The organizations that solve this stop treating it as a remediation problem.

Step one: count, then triage

You can't plan against an unknown quantity, and almost no organization knows theirs. Crawl the site and count.

Then sort everything into three piles:

• Active. Downloaded regularly, or required to access a service. Usually a startlingly small fraction — often a few percent of the library.
• Archival. Retained for records or legal reasons, rarely accessed.
• Obsolete. Superseded, expired, or duplicated. In most libraries this is a large share.

This step alone routinely removes a third or more of the problem. A document nobody has opened in six years and that no retention schedule requires does not need remediation. It needs to be retired.

Step two: stop making PDFs of things that should be pages

For the active pile, the reflexive answer is remediation. Often the better answer is conversion.

A permit application, a fee schedule, a contact list, a policy — these are usually more usable, more searchable, and cheaper to maintain as HTML pages than as remediated PDFs. An accessible web form beats an accessible PDF form for nearly everyone, including people with no disability at all.

The test: does this document need to be a document? Does anyone print it, sign it, or file it? If not, it probably wants to be a page.

Step three: remediate what genuinely must stay

Some things really are documents. Signed resolutions, formal reports, records with legal standing. Those get real remediation, prioritized by use.

Budget honestly here. A born-digital document exported from a well-structured source is comparatively quick. A scanned image requires OCR first, and the result needs review — OCR confidently misreads numbers, and a budget table with wrong figures is worse than an inaccessible one.

Step four: fix the source, or you're doing this again

This is the step that determines whether the effort was worth it.

Documents come from somewhere: Word templates, InDesign files, a reporting tool, an agenda management system. If those produce untagged output, remediation is a permanent tax.

Fix the templates so heading styles, alt text prompts, and table headers are built in. Train the people who produce documents — it's typically a couple of hours of training, and it changes the trajectory of the entire backlog. Configure the export settings on your agenda and reporting systems. Put an accessibility check in the publishing workflow with a named owner.

Do that and the backlog becomes finite. Skip it and no amount of remediation budget ever closes it.

The sequence, compressed

Count it. Retire what's dead. Convert what should be a page. Remediate what must stay a document, by priority. Fix the templates and train the authors so the library stops growing.

That's a plan with an end. "Start remediating PDFs" is not.