Should this be a PDF at all? When HTML beats remediating a PDF

Ask accessibility practitioners how to fix a library of PDFs, and the first answer is often a question: why is this a PDF? For a real share of most libraries, that is the right question, and the honest answer is that the document should have been a web page. For the rest, a web page is not an option, because the PDF is the record. This guide helps you tell the two apart, and shows what you can take off the list for free, before you spend anything. Ockem sells PDF remediation, so we are not neutral here. Two sections of this guide argue against buying it: one for content you are still writing, and one for content you can lawfully archive. As of October 2026.

Three questions that sort a library

1. Is someone writing or rewriting this content now? Then publish it as a web page. Don't pay to remediate a document that isn't finished, and don't pay again next year for the next edition. Fix the template instead.
2. Is the PDF the record? The form that was filed, the minutes as adopted, the report other people cite by its address, the scan of a paper original. Then the PDF itself has to be made accessible, because there is no web version to publish in its place, only a copy you would have to keep in step with the record.
3. Is it inactive? Superseded, kept for reference, rarely opened. Then the ADA Title II rule may already exempt it at no cost, if you meet the archive conditions as written.

Only one of the three piles has a price. The costly mistake is not choosing the wrong format. It is pricing the whole library before you sort it.

When HTML is the better answer

When this applies, it is not a close call. It helps to be precise about why, because the usual shorthand, "PDFs are inaccessible," is not quite true, and a vendor can easily wave it away.

A tagged PDF can be read, but its layout is still fixed

Tags give assistive technology what it needs to interpret a page: a reading order, real headings, alt text on images, table structure and a declared language. That work is real, and it is most of what remediation is.

A PDF page is still a fixed layout. WCAG 2.1 success criterion 1.4.10 Reflow asks that content work at a width equal to 320 CSS pixels without scrolling in two directions, which W3C equates to a 1280-pixel-wide window zoomed to 400%. A web page does that on its own. A PDF does it only through the reader: Acrobat's Reflow view and some mobile apps rebuild a well-tagged document as a single column, and they depend on the tags being right. That is also how Ockem Proof scores Reflow: it passes only when the tagging, reading order and images-of-text checks pass. A pass means the file can be reflowed by a reader that supports it. It doesn't mean every reader a person uses will do it well.

Criterion 1.4.12 Text Spacing is written for content in markup languages that let a reader override line height, letter spacing and word spacing. PDF readers don't offer that override, so W3C's guidance for non-web documents treats it as a criterion a PDF can neither meet nor fail. No remediation adds it.

Both are properties of the format, not defects in one file, so they survive a perfect remediation. A well-tagged 20-page report is still a 20-page fixed-layout report to someone reading it at high magnification on a phone. When people who read that way say they would rather get a web page, they are describing the format, not your tags.

A template fix is paid once, and a PDF fix is paid per file

Fix the heading structure in a page template, and every page built from it inherits the fix, including pages that don't exist yet. Fix it in a PDF, and you have fixed one PDF. If the document comes out every month, you have signed up to pay for the same fix every month. This is the strongest argument for HTML, and it gets stronger the more often you publish. A web page also needs no download, can be linked at the exact heading someone needs, and can be corrected without chasing every copy of a file.

"Default to HTML" is federal policy, for federal agencies

OMB Memorandum M-23-22, Delivering a Digital-First Public Experience, tells federal agencies to default to HTML for web content rather than formats built for printing or for preserving a layout, such as PDF, unless a specific user need calls for another format. Read its scope before you cite it. M-23-22 binds federal agencies. The ADA Title II rule for state and local governments sets a WCAG 2.1 AA standard and does not require any particular format. For a city or county, "default to HTML" is good guidance rather than an obligation, so the decision has to be made document by document.

If you're writing it now, write it as a web page

If a report goes out next month and its layout isn't essential, the accessible version of that report is a web page. Paying to remediate it means paying for the same work twice: once to lay it out as a document, and again to explain the layout to a machine. Don't send us next year's edition either. The remediation program worth having is the one that shrinks every year because fewer PDFs are being made.

When there is no web version to publish instead

Most existing libraries can't follow that advice, because nobody chose the documents in them. They piled up. Four situations come up again and again, and in none of them is "publish HTML instead" a real option.

The PDF is the record

A form that was filed. Minutes, ordinances and resolutions as adopted. A signed publication. A scan of a paper original in a filing cabinet. In each case the PDF is not one way of showing the record. It is the record, and a retyped web page is a copy. A copy may be fine for a reader, but it usually isn't fine for whoever is accountable for the record, so you end up maintaining two versions and reconciling them for good. The barrier is in the PDF, so the fix has to be in the PDF.

The address is already in use

A document that has been on your site for years is linked from other agencies, cited in filings and indexed by search engines. Replacing it means breaking those links, or keeping a redirect and two versions. Posting an accessible web version beside a PDF that stays posted and linked leaves the PDF exactly as reachable, and exactly as much of a barrier, as before. If it stays up, it is still your content.

A system you don't control only exports PDF

Agenda software, permitting platforms, tax and assessment systems, state-mandated forms and GIS exports often produce PDF and nothing else, from templates the vendor owns. "Publish HTML instead" is advice for your next contract renewal, not a plan for the documents the system produced last week. Raise it at renewal anyway, and ask in writing for the underlying data or an HTML output.

The layout is the content

Maps, plan sets, engineering drawings and forms whose arrangement carries meaning are usually poor candidates for HTML and hard cases for remediation. What they need is a text equivalent written by someone who understands the content: what the map shows, what the drawing specifies. No format choice and no software removes that task. WCAG 1.4.10 exempts content that needs a two-dimensional layout for its meaning, such as maps and diagrams, but that exemption covers reflow only. You still have to communicate what the drawing says.

Converting everything is a bigger project than it sounds

An automatic PDF-to-HTML conversion carries the original's problems across. Headings that were only bold text stay bold text, tables that were only aligned text stay aligned text, and images of text stay images. You do the same structural work without a page-faithful original to check against, and you end up with two versions instead of one. If content deserves to be a web page, it usually deserves to be rewritten as one.

Archive first: the step that costs nothing

Before you price anything, find out how much of the library you don't have to fix. Experienced practitioners recommend this first on a large legacy library, and it costs nothing.

This section covers the DOJ's ADA Title II web rule for state and local governments. As of October 2026, its compliance dates are April 26, 2027 for governments serving 50,000 or more people and April 26, 2028 for smaller governments and special district governments. The rule defines conventional electronic documents to include PDFs, so a PDF on your site is covered unless an exception applies.

The four archive conditions

Under 28 CFR 35.104, archived web content is exempt only if all four of these are true.

1. It was created before your compliance date, or it reproduces paper or other physical media created before that date.
2. It is kept only for reference, research or recordkeeping.
3. It has not been changed since it was archived.
4. It sits in a dedicated area or areas clearly identified as archived.

Two of the four do most of the failing. Condition 2 fails for a document still linked from a live service page, because people sent to it to get something done are not using it only for reference. Condition 4 fails often, because old files tend to sit wherever they were uploaded, beside current ones, with nothing marking them as archived. Meeting it takes an archive section labeled as one and a record of what moved there and when, which still costs far less than remediating what moves there.

A separate exception in 28 CFR 35.201(b) covers PDFs posted before your compliance date, unless people currently use them to apply for, gain access to or take part in a service. A 2019 permit application that people still download and file is not exempt because of its age. Confirm with counsel how the new compliance dates apply to your documents. Our guide to archive, keep or fix has a decision table by document type and the steps for setting up an archive area.

An exception is not a waiver

You still owe effective communication under 28 CFR 35.160 and reasonable modifications under 35.130(b)(7). DOJ's compliance guide says that even when content doesn't have to meet WCAG 2.1 AA, you "would likely still need to provide the content to a person with a disability who needs it in a format that is accessible to them."

So an exception means you don't have to fix the content in advance. You still need a way to turn a request into an accessible copy in time. If your archive holds 10,000 documents and requests arrive a few times a year, that is cheap and you have saved a lot. If they arrive every week, the content was not really archival.

Keeping a record and publishing it are separate decisions. A records system can hold a record for its full retention period after the web copy comes down, though some state posting laws require certain documents to stay online for a set time.

Sorting a library in six steps

1. Count first. List every PDF with its address, page count, date and what links to it. Our guide to finding every PDF covers the methods.
2. Sort by use, not by folder. Ask whether people need the document to apply for, access or take part in something, and whether anyone opens it.
3. Archive or remove what is inactive, and record what you did and when.
4. Delete the duplicates, such as a flyer that repeats a web page.
5. Fix the sources that produce most of your new PDFs, or the backlog will refill.
6. Price what is left. That number is the size of your remediation project.

What usually lands where

These are typical answers. Your facts decide.

Where Ockem Proof fits

This is the part where we sell something, marked so you can weigh the rest of the guide without wondering.

Ockem Proof fixes and certifies the pile that is left after sorting: the records, the filed forms, the scanned archive and the documents already cited by their address. Importing your documents and running the first check on each one is free, and the free part tells you how big the paid part is. Proof charges per document, from $59 for a single document to $15 each in a 1,000-document pack, and you pay only when a document is certified. Every fix is re-checked automatically before it counts, and your team answers the judgment calls software shouldn't make. Proof evaluates every WCAG 2.1 A/AA criterion that applies to a document, by machine check or recorded attestation, and the certificate gives the count, every one of them. If a criterion that applies was not evaluated, the certificate names it. Our coverage page shows how each criterion is evaluated.

What we won't do is price your whole domain. If the sorting above empties most of your library, that is a good outcome, and the right thing for us to tell you is that you need less than you thought.

Where to start

Start with a count. Our free PDF inventory counts the PDFs on your site and includes a scorecard on a sample of your documents (12 by default, up to 100 for requests we qualify), usually within 3 business days. For the full rule and its exceptions, read Title II and PDFs, and to compare pricing models, see our cost guide. To see what Proof does with one of your records, you can try one PDF (opens an external site) of up to 10 pages free, with no account, only a verified email.

Questions

This guide explains the rules in plain language. It isn’t legal advice. Check how they apply to your organization with counsel.