PDF Accessibility

PDF Accessibility Audit: How to Review a Large Document Set

A repeatable PDF accessibility audit workflow for large document collections: inventory files, triage risk, separate automated checks from human review, and preserve evidence of remediation.

People in the office

A pdf accessibility audit for a large document set is not one checker run and a spreadsheet of scores. It is a repeatable process: inventory every file, triage risk, run consistent technical checks, separate automated findings from human review, remediate, recheck, and keep evidence of what changed.

The hard part is not checking one PDF. It is keeping control of hundreds or thousands of files with different owners, templates, versions, audiences, and publication risks.

Treat the audit as a portfolio review. The deliverable should help a team answer three questions: which PDFs exist, which ones matter most, and what evidence shows they were checked, repaired, reviewed, or removed.

Updated November 14, 2026.

What a PDF accessibility audit should produce

A large-set PDF accessibility audit should produce more than a pass/fail result. It should produce a working record of the collection and a defensible remediation plan.

At minimum, the audit should leave you with:

  • A current document inventory showing every PDF in scope and where it lives.

  • Findings by file and issue type, not only by overall score.

  • Remediation priorities based on audience, usage, risk, and operational importance.

  • Status fields showing whether each file is unchecked, checked, repaired, rechecked, removed, replaced, or awaiting human review.

  • Ownership records showing who can approve edits, supply source files, or retire the document.

  • Evidence reports showing what was checked, what changed, and what still needs human judgment.

This is the main difference between a single-PDF accessibility check and a portfolio audit. A single file can be opened, tested, repaired, and reviewed in isolation. A collection needs scope control, triage, version tracking, publication decisions, and repeatable evidence.

A collection audit should not collapse everything into one accessibility score. A score hides the work that matters: a public benefits application with broken form fields is not equivalent to an outdated internal flyer with missing bookmarks.

Start with a complete inventory of the PDF collection

The first failure in large PDF audits is usually scope. Teams start testing the files they know about, then later discover another repository, course page, policy library, intranet folder, or public download archive.

Start by finding every PDF in scope. That includes PDFs linked from:

  • Public websites

  • Department pages

  • Student portals or learning management systems

  • Knowledge bases

  • Document libraries

  • Staff intranets

  • Download folders

  • Forms pages

  • Archived web pages that are still reachable

  • Email-linked or campaign-linked landing pages

NYU’s PDF content audit guidance makes the same basic point: take inventory of the PDF documents linked from the public website before trying to manage compliance work at scale (NYU).

A website PDF scan can help find PDFs before detailed review begins. It does not replace the audit register; it feeds it.

For each file, record the fields that let someone make a decision later:

  • URL or storage location

  • Document owner

  • Title

  • Document type, such as policy, form, manual, report, syllabus, brochure, journal article, thesis, or financial notice

  • Age or publication date

  • Audience

  • Language

  • Page count

  • Update frequency

  • Publication status

  • Whether the file is still needed

Then add operational fields:

  • Last reviewed date

  • Current version

  • Source file location, if known

  • Replacement HTML page, if available

  • Public-facing or restricted

  • Department or business unit

  • Accessibility review status

  • Remediation owner

  • Next review date

PDF collection inventory flow from document sources to an accessibility audit register

Flag files that will need special handling:

  • Duplicates: choose the canonical file before repairing copies.

  • Superseded versions: retire or redirect them if they are no longer authoritative.

  • Broken links: fix publication hygiene before testing.

  • Scanned files: these may lack a usable text layer.

  • Forms: these need field labels, instructions, keyboard behavior, and error handling checks.

  • Equations: these may need math-specific review.

  • Unknown owners: these slow approval and remediation.

  • Missing source files: these may require direct PDF repair instead of re-export from Word, InDesign, LaTeX, or another authoring tool.

If the collection is large, use the inventory as the source of truth. A document-management process matters here as much as an accessibility process. If the team already struggles with naming, duplication, and version control, general document management practices should be fixed alongside the audit workflow.

Set the audit scope and triage documents by risk

Do not promise that the audit covers “all PDFs” unless you can define where “all” ends.

A useful scope statement says what is included and excluded. For example:

  • All active public PDFs on the main website

  • All PDFs in a benefits application journey

  • All current course handbooks for one academic year

  • All patient-facing PDFs in a healthcare portal

  • A representative sample from each document template

  • All PDFs published or updated after a certain date

  • One department, website section, or document library

Record exclusions and the reason for each. “Archived and no longer linked from public pages” is different from “unknown owner” or “outside current remediation budget.”

Triage should decide review order, not accessibility status. A file labelled “routine” has not been found accessible. It has only been placed behind higher-impact work.

Useful triage labels are simple:

  • Urgent: public, high-use, legally or operationally important, or blocking access to a service.

  • High: public-facing, current, important to a known audience, or frequently downloaded.

  • Routine: active but lower-risk.

  • Archive: retained for historical or records reasons, but not part of a current user journey.

  • Remove: obsolete, duplicated, superseded, or no longer needed.

Prioritize documents that are:

  • Public-facing

  • Frequently used

  • Recently updated

  • Legally or operationally important

  • Required for applications, benefits, enrollment, employment, healthcare, billing, safety, or academic progress

  • Used by students, patients, applicants, customers, staff, or the public

A 200-page internal archive may take longer to repair than a two-page public form, but the form may matter more. Audit sequencing should reflect impact, not just page count.

For the narrower question of archive cleanup, use a dedicated archive plan rather than turning the whole audit into a records-disposition project. Ordecto’s guide to prioritizing legacy PDF archives for accessibility remediation covers that decision in more depth.

Separate machine-checkable findings from human review

A large audit needs a clear line between what software can check consistently and what a person must judge.

A tag is structural markup inside a PDF. Tags identify elements such as headings, paragraphs, lists, tables, figures, links, and form fields. Section508.gov describes tags as the structural foundation of an accessible PDF because they tell assistive technology what each piece of content is (Section508.gov).

A tagged PDF is not automatically a good PDF. A heading can be tagged as a paragraph. A table can have cells but no usable header relationships. A figure can have a description that is technically present but wrong.

Reading order is the sequence in which assistive technology encounters the content. It matters because many PDFs are visually arranged in columns, sidebars, footnotes, callouts, headers, captions, and floating objects. The visual layout may look fine while the assistive-technology order jumps from a sidebar into a footnote and back into the main sentence.

The Commonwealth of Massachusetts accessibility training explains that the tags panel displays the structure most assistive technology uses to understand a PDF, which is why tag order and structure matter beyond visual appearance (Mass.gov).

Alt text is a concise text description of an informative image. A checker can detect whether a description field is empty. It cannot reliably decide what the image means in context. A campus photo used as decoration does not need the same treatment as a chart showing graduation outcomes or a diagram explaining medication dosage.

Automated PDF accessibility checks compared with human review tasks

For each issue, record one of four statuses:

  • Checked: the item was tested and no issue was found within the audit method.

  • Failed: the item was tested and an issue was found.

  • Needs a person: the item cannot be judged by automation alone.

  • Not tested: the audit method did not cover it.

That distinction prevents a common reporting error: presenting “no automated failures found” as “the PDF is accessible.” Those are not the same claim.

For the deeper limits of automated testing, see What Automated PDF Accessibility Checkers Cannot Test.

Apply the same technical review to every in-scope file

Consistency matters more than elegance. If different reviewers use different criteria, the collection report becomes hard to trust.

Use the same technical review categories for every in-scope file, then add document-specific tests when needed.

Start with basic document properties:

  • Document title: does the metadata identify the document clearly?

  • Document language: is the primary language set correctly?

  • File name and visible title: are they understandable and aligned?

  • Security settings: do they interfere with assistive technology or extraction where review requires it?

Then check whether the file has a usable text layer. A scanned PDF may look readable to a sighted person while containing only page images. Optical character recognition, or OCR, converts scanned text into selectable text that assistive technology can access. For scans, do not stop at “OCR exists.” Verify recognition quality on representative pages, especially pages with columns, stamps, handwriting, tables, or poor contrast.

Section508.gov’s PDF testing training includes manual evaluation as part of PDF accessibility testing, which is the right model for large audits: tools help, but reviewers still need a method (Section508.gov).

Review the structure:

  • Heading levels and hierarchy

  • Paragraph tagging

  • List tagging and nesting

  • Table headers and relationships

  • Link tags and link text

  • Form fields, labels, instructions, and error cues

  • Figures and descriptions

  • Decorative content marked as artifacts, meaning content assistive technology can ignore

  • Bookmarks for longer documents

  • Logical reading order

  • Footnotes, captions, sidebars, and multi-column flows

Include keyboard access where relevant. This is especially important for interactive forms, buttons, checkboxes, radio buttons, drop-downs, and any workflow where a person must complete or submit information.

Check contrast and other applicable requirements according to the document type and evaluation scope. A policy PDF, a form, a scientific report, and a marketing brochure will not have identical risks.

For each finding, preserve:

  • File version

  • Date checked

  • Reviewer

  • Test method

  • Affected page or page range

  • Issue type

  • Severity or priority

  • Remediation owner

  • Status

  • Notes on human-review items

Do not reduce the collection to one score. A portfolio-level audit should let stakeholders filter by owner, audience, risk, issue type, and status.

Use PDF/UA and WCAG as review references without treating them as certificates

PDF/UA is the PDF-specific accessibility standard. WCAG, the Web Content Accessibility Guidelines, is a broader accessibility framework used for web and digital content. They address related goals, but they are not interchangeable labels for one automated test.

PDF/UA focuses on what a PDF file must contain so assistive technologies can interpret its structure. WCAG is broader: it addresses whether digital content is perceivable, operable, understandable, and robust across formats and user needs.

For audit governance, record:

  • Which standard versions the audit references

  • Which evaluation methods were used

  • Which tools were used

  • Which manual checks were performed

  • Which document types were excluded or sampled

  • Which organizational or jurisdictional requirements apply

This is not legal advice, and an audit report is not proof that an organization has met every regulatory obligation. It is evidence of a review process and the state of the files under that process.

Where a tool is useful for consistency, use it for the repeatable parts. Ordecto’s PDF/UA checker, a PDF accessibility tool from the team behind Otio, checks every page against PDF/UA-1 with WCAG 2.2 as a referenced test model and reports findings as failed, needs a person, or not tested. Its free PDF accessibility checker can help with initial checks before a file enters remediation.

That result is still not a certificate. No automated checker can independently judge every reading-order decision, image meaning, complex table relationship, or screen-reader experience.

Run remediation as a repeatable collection workflow

Once the audit finds issues, remediation should run as a controlled loop, not as one-off file fixing.

Start with sampling when documents share templates, authors, or production processes. For example, if 120 departmental policy PDFs were exported from the same Word template, checking a representative sample can reveal the template defects before the team repairs every file individually.

Use full-collection review when:

  • The files vary widely

  • The documents are high-risk

  • The publication requirement covers every file

  • The collection includes many forms, scans, tables, or complex layouts

  • The sample reveals inconsistent production quality

Group recurring findings across files. Common patterns include:

  • Missing or incorrect headings

  • Lists tagged as plain paragraphs

  • Tables without headers

  • Links with vague text such as “click here”

  • Images without meaningful descriptions

  • Decorative lines, icons, or background graphics exposed to assistive technology

  • Broken reading order in multi-column pages

  • Scanned files with poor OCR

  • Form fields without labels

  • Missing document language or title

This is where portfolio auditing earns its keep. If 70 files share the same inaccessible template, repairing the template or export process prevents repeated defects. If every new syllabus is exported without headings, the fix belongs in author training or the source-document template, not only in the PDF repair queue.

PDF accessibility audit and remediation loop

Use PDF remediation for repairs such as headings, lists, tables, links, figures with descriptions, form fields, reading order, and text layers. Treat the repaired PDF as a candidate for review, not as a certified final state.

After repair, a person should review the areas where automation is weakest:

  • Does the reading order make sense when followed linearly?

  • Does the image description convey the information the reader needs?

  • Are complex tables understandable with headers and relationships?

  • Do form fields announce useful labels and instructions?

  • Can keyboard users reach and operate interactive content?

  • Does a screen reader expose the document in a usable sequence?

Then recheck the repaired file. Compare it with the original. Update the inventory. Keep the repaired file, the report, the reviewer, the date, the findings, and unresolved items.

For the narrower evidence-record problem, see Building an Audit Trail for PDF Accessibility Remediation. If the next question is resourcing rather than evidence, use a separate effort estimate; Ordecto’s guide to estimating PDF accessibility remediation effort is the better place for that.

This workflow also benefits from ordinary document operations. If your organization already uses document routing, approvals, and recurring updates, connect the accessibility status to that system rather than keeping a disconnected spreadsheet. A broader document workflow automation process can help prevent newly published PDFs from bypassing review.

Publish an inventory and evidence report that stays current

The final audit output should help stakeholders act. It should not be a static compliance-looking artifact that no one updates.

Give stakeholders a collection-level view with:

  • Document owner

  • Audience

  • Priority

  • Current version

  • Audit status

  • Remediation status

  • Next review date

  • Unresolved human-review items

  • Removal or replacement decision

  • Notes on files awaiting source documents or owner approval

Keep separate fields for automated findings and human observations. A file with no machine-detected failures should not be presented as fully accessible if reading order, alt text meaning, complex tables, or screen-reader behavior still need review.

Record removals and replacements as decisions. Do not silently delete files from the audit universe. Note when a PDF was:

  • No longer needed

  • Replaced by an accessible HTML page

  • Superseded by a newer file

  • Retained for archive reasons

  • Retained with a documented review plan

  • Redirected to a canonical version

Finally, define a maintenance trigger. A large audit loses value if new PDFs bypass the same process.

Triggers can include:

  • New public PDF publication

  • Updated PDF upload

  • New form release

  • Annual policy refresh

  • Course-material rollover

  • Website migration

  • Brand-template change

  • Source-document template change

  • Departmental handoff

  • Procurement of PDF-producing software

The maintenance rule should be simple: no new or updated PDF enters the collection without inventory, triage, technical check, human review where needed, remediation status, and evidence.

Run Ordecto’s website PDF scan to find the PDFs in your collection and identify which documents should enter detailed accessibility review.

FAQ

Q: Can an automated PDF checker complete a large document-set audit?
A: No. It can make repeated technical checks more consistent and identify files that need attention, but it cannot replace human review of meaning, reading order, complex tables, or screen-reader behavior. A portfolio audit also needs inventory, triage, version records, ownership, and evidence.

Q: What should be recorded for each PDF in an accessibility audit?
A: Record its location or URL, owner, title, type, age, audience, language, page count, update frequency, version, review date, findings, reviewer, remediation status, and unresolved human-review items.

Q: Should every PDF receive the same level of review?
A: No. Start with public-facing, frequently used, recently updated, legally or operationally important, and high-impact documents. Document the scope and triage method so lower-priority files are not forgotten.

Q: Does a PDF/UA or WCAG check prove that a PDF conforms?
A: No. PDF/UA and WCAG provide useful accessibility reference points, but an automated result is not a certificate or proof of conformance. Human evaluation and the applicable organizational or regulatory context still matter.