PDF Remediation Quality Control: What Should You Check Before Publishing an Accessible PDF?

You’ve remediated your PDF. Tags are added, alt text is written, and reading order looks tidy in the HTML preview or tag tree panel. 

But accessible content doesn’t end with remediation; remediation and verification are two different jobs. Skipping verification is how “accessible” PDFs end up back in someone’s inbox with a complaint attached. A tag tree that looks clean in your authoring tool can still fail the moment a real screen reader touches it. The only way to know for sure is to check it yourself, in a few different ways.

This is your QC checklist. These are specific things to check, why they matter, and how to test them.

How do I know if my PDF is actually accessible?

Short answer: no single tool can tell you that on its own.

Automated checkers like Adobe’s Accessibility Checker, PAC (PDF Accessibility Checker), or Equidox’s validation tools are excellent at catching structural problems: missing tags, untagged content, missing document language, absent titles. What they can’t catch is meaning. A checker will approve an image tagged with alt text that says “image47.jpg” because, technically, alt text exists. It also won’t understand that your table headers are associated with the wrong cells if the markup is technically valid but logically wrong. It won’t know whether your reading order makes sense.

So “actually accessible” requires two passes:

1. Automated validation to catch the mechanical failures fast, like missing tags, unset document properties, contrast ratios below threshold, unstructured lists.
2. Manual verification, ideally including using an actual screen reader (NVDA and VoiceOver are both free), to confirm the document makes sense when experienced non-visually, not just when scanned by software.

If you only do step one, you have a document that passes a test. If you do both, you have a document that works for a person.

The pre-publish checklist

Here’s what to walk through before you consider a remediated PDF done. Think of it less as a formality and more as your last chance to catch something before a user does.

1. Document properties and metadata

It’s easy to skip over these elements,but they’re also the first things screen readers announce, so they matter more than you think.

• Set a descriptive title: Set in document properties, not just the filename. “Q3_Report_FINAL_v3.pdf” is not a title.
• Set the language: Declared at the document level (and overridden at the paragraph level for mixed-language content). Without it, screen readers guess the pronunciation engine, often badly.
• Flag the PDF as tagged: Confirm the document is actually marked as tagged in its properties, not just visually tagged in your editor.

2. The tag tree: structure, not just presence

Having tags isn’t the goal. Having the right tags in the right order is.

• Check that every piece of content has a tag; no orphaned or untagged elements sneaking through, especially in complex layouts with sidebars, pull quotes, or multi-column text.
• Check whether heading levels are logical and sequential (H1 → H2 → H3, no skipping from H1 to H4 because it “looks right” visually).
• Determine if decorative elements (background graphics, dividers, watermarks) are marked as artifacts so screen readers skip them instead of announcing “image, image, image” through your header.
• Ensure that reading order matches visual order, especially in multi-column layouts, sidebars, and pull quotes. This is the most common failure point in remediated PDFs, because visual layout tools don’t inherently understand reading sequence.

3. Alternative text: quality, not just quantity

This is where a lot of remediation projects fail their real purpose.

• Check that every meaningful image has alt text that describes its function or content, not its filename.
• Mark purely decorative images as artifacts. Forcing alt text onto decoration is unnecessary.
• Determine if complex images like charts, graphs, or infographics have a full-length description nearby in the body text, because a single alt attribute can’t carry a chart’s worth of information.
• Create alt text that describes what someone needs to know, not just what the image looks like. “Bar chart” is a caption. “Bar chart showing revenue grew 18% year-over-year, driven mostly by Q4” is alt text.

4. Tables

Tables are where automated checkers and human reviewers most often disagree, because a table can be technically tagged and still be functionally unusable.

• Ensure header cells (<TH>) are correctly marked and scoped (row, column, or both) and not just formatted to look like headers.
• Use Headers/ID attributes for complex tables with merged cells or multi-level headers  so each data cell is explicitly associated with its header. Scope alone often can’t resolve the relationship.
• Avoid tables used purely for visual layout (a holdover from print-design habits) or that are not marked as data tables.

5. Forms

If your PDF has fillable fields, this section alone deserves its own QC pass.

• Include a programmatically associated label for every field, not just placeholder text that disappears the moment someone tabs into the field.
• Ensure tags are arranged in a logical sequence, matching the visual flow a sighted user would follow.
• Communicate required fields, format expectations (dates, phone numbers), and error states in text, not by color or symbol alone.
• Use meaningful field type: checkboxes are checkboxes, radio buttons are radio buttons and not visually styled text boxes standing in for interactive controls.

6. Color and contrast

Easy to overlook because it’s often unchanged from the source design, but it is still your responsibility to verify that there is sufficient contrast for the content to be readable.

• Ensure text meets WCAG contrast minimums. Use 4.5:1 for normal text, 3:1 for large text.
• Convey information by other ways than color (for example, a red cell in a table meaning “over budget” needs a text label or symbol too).
• Distinguish links from surrounding text by more than color.

7. Bookmarks and navigation

For longer documents, bookmarks are the equivalent of a table of contents for someone navigating by keyboard or screen reader.

• Confirm bookmarks exist for documents over roughly 10–15 pages and mirror the heading structure.
• Ensure bookmarks are nested correctly, matching the logical hierarchy of the content.

What should I check after remediating a PDF?

If you want the compressed version of everything above, here’s the sequence:

1. Run an automated check (Acrobat’s checker or PAC) to catch structural gaps. Equidox Software also flags untagged elements before export.
2. Open the tag tree manually and read it top to bottom. Does the order match what a sighted reader would experience? If you have Equidox Software, open the HTML preview and ensure that what you’re seeing matches the PDF.
3. Turn on a screen reader and navigate the actual document. Listen for anything that sounds wrong, out of order, or missing.
4. Tab through interactive elements, such as forms and links, using only the keyboard. If you can’t reach something or the tab order seems incorrect, that’s a real user’s experience too.
5. Spot-check alt text by reading it in isolation, without looking at the image. Does it still make sense?
6. Verify metadata: check the title, language, tagged status.
7. Check contrast on any text or elements that weren’t already validated in the source design.

How do you QA an accessible PDF, really?

The honest answer is that QA for accessibility isn’t a checklist you run once. It’s a habit you build into your workflow, the same way Spellcheck became automatic instead of optional. The documents that consistently pass review undergo a repeatable process: automated scan, manual tag review, assistive technology test, sign-off.

That repeatability is exactly the problem Equidox was built to solve. Manual remediation is where accessibility work usually breaks down at scale. Doing this checklist by hand, on every document, every time, doesn’t hold up against real deadlines and real document volume. Equidox’s remediation platform builds structural accuracy, like correct tagging, reading order, table structure, into the process itself, so your QC pass is confirming good work instead of discovering gaps you have to fix under time pressure.

The bottom line

An accessible PDF isn’t one that passes a scanner. It’s one that makes sense to a person who’s never seen it. The checklist above is how you find out, before your users do, whether the remediated PDF is compliant and usable. While automated tools are helpful to identify missing tags, real accessibility needs to be confirmed by a human being.

Build this check into every PDF publication workflow, and “accessible” stops being a label you apply and starts being something you can actually stand behind.

Nina Overdorff

Nina comes to Equidox with years of sales and marketing experience from a variety of industries and holds a BS in Language Arts Education. Nina has a passion for words, storytelling, and information, which she believes everyone should have access to regardless of ability. After spending time as a teacher with a blind student, she became much more aware of the limitations and abilities of web accessibility, and how essential it is to those experiencing disabilities. “Being able to access information equally ensures that everyone has an equal opportunity for education, employment, and success in life.”