Summit 12 October 2023 📍 Online

PDF Accessibility Summit 2023 — Practical PDF/UA Implementation

The PDF Accessibility Summit 2023 was a focused one-day virtual workshop hosted by the PDF Association on 12 October 2023, dedicated entirely to the practical challenges of implementing PDF/UA accessibility requirements. Sessions covered tagging workflows in authoring tools, remediation of legacy documents, automated testing with PAC 2024 and veraPDF, and legal compliance obligations under ADA, Section 508, and the approaching European Accessibility Act deadline.

Official event page →

Key takeaways

  • The most common PDF accessibility failures are: missing heading structure, incorrect multi-column reading order, unmarked decorative images, and tables without header cells.
  • PDF/UA conformance checking uses the Matterhorn Protocol — 136 checkpoints, split between machine-verifiable (PAC/veraPDF) and human-verifiable.
  • Fix at source (Word/InDesign export settings) is more scalable than post-export PDF remediation for regularly produced documents.
  • Automated tool validation must be supplemented with manual screen reader testing — PAC-passing PDFs can still fail in practice.
  • For large backlogs, categorise by user impact and frequency, automate detection, prioritise manual remediation for high-impact documents.
The PDF Accessibility Summit 2023, held online on 12 October 2023, was a single-day event organised by the PDF Association focused exclusively on practical PDF accessibility implementation. Unlike the broader PDF Days conferences, the Summit addressed a specific practitioner audience: accessibility specialists, content authors, and document engineers responsible for producing and remediating accessible PDFs in real organisations.

PDF/UA implementation in practice

PDF/UA (ISO 14289) defines the accessibility requirements for PDF documents. The Summit's practical focus examined the gap between what the specification requires and what organisations actually deliver. The Matterhorn Protocol — published by the PDF Association alongside PDF/UA — defines 136 checkpoints that characterise PDF/UA conformance, dividing them into: - Machine-verifiable checkpoints (can be checked by automated tools like PAC or veraPDF) - Human-verifiable checkpoints (require human judgement — e.g., whether alt text accurately describes an image, whether reading order is logical) Summit sessions revealed consistent failure patterns across organisations: 1. **Missing or incorrect heading structure** — The most common failure. Documents exported as flat PDFs with no heading tags, or with headings applied visually (bold text at large size) but not tagged as H1/H2/H3 in the PDF structure tree. 2. **Incorrect reading order in multi-column layouts** — PDF generation tools often produce PDFs where the text extraction order follows document page order rather than the intended reading order, causing screen readers to read text from the wrong column. 3. **Decorative images without /Artifact marking** — Images that are purely decorative (borders, backgrounds) should be marked as /Artifact so screen readers skip them. Unmarked decorative images generate spurious alt text announcements. 4. **Tables without header cells** — Data tables require /TH header cells with /Scope attributes; presentational tables require /Role set to /Presentation.

Remediation workflows and tooling

The Summit covered the two main approaches to producing accessible PDFs: Fix at source — Correct the accessibility problems in the source document (Word, PowerPoint, InDesign) and re-export. This is the most scalable approach for regularly produced documents. Microsoft Word's Accessibility Checker (built into the Review tab) identifies most heading, alt text, and table header issues before export. Adobe InDesign has a dedicated Export to Tagged PDF workflow with accessibility options. Remediate the PDF directly — When source documents are unavailable (scanned archives, third-party documents), use Adobe Acrobat Pro's Accessibility tools or specialist tools like CommonLook PDF to add and correct the tag tree directly. This is labour-intensive but necessary for legacy document backlogs. For organisations with large backlogs, the Summit recommended a tiered approach: 1. Categorise documents by usage frequency and user impact 2. Apply automated fixes where possible (PAC or veraPDF in batch mode to identify and flag failures) 3. Prioritise high-impact, high-frequency documents for manual remediation 4. Create accessible templates and authoring guidance to prevent new accessibility failures in future production Adobe Acrobat's accessibility features provide the most complete remediation toolset, but licensing costs motivate many organisations to explore open-source alternatives for validation (veraPDF) and bulk processing.

Assistive technology testing

A critical track at the Summit addressed the gap between automated tool validation and actual user experience. A PDF can pass all 136 Matterhorn Protocol checkpoints and still be difficult to use with specific screen readers — because different AT (assistive technology) software interprets PDF structure differently. Primary testing combinations for US and EU markets: - **NVDA + Chrome** — Most common free screen reader setup; widely used by screen reader users - **JAWS + Edge/Chrome** — Most common enterprise and workplace screen reader - **VoiceOver + Safari (macOS/iOS)** — Default Apple screen reader - **TalkBack (Android)** — Default Android screen reader, used with Chrome Summit sessions demonstrated that the same PDF can behave differently across these combinations — particularly for: - Complex table navigation - Form field announcement and focus management - Figure/caption association - Document navigation via headings (using the AT's heading navigation shortcut) The recommended practice is to validate with PAC/veraPDF first (catch all machine-verifiable failures), then conduct manual testing with at least two screen reader / browser combinations (NVDA+Chrome and JAWS+Edge are the recommended minimum for US/EU markets). The WebAIM Screen Reader User Survey provides current data on screen reader market share to inform testing prioritisation.

Frequently asked questions

What is PDF/UA?

PDF/UA (ISO 14289) is the ISO standard for accessible PDF documents. It specifies requirements for structure tags, reading order, alternative text, and form field accessibility. A PDF/UA-conformant document can be navigated and understood by users of screen readers and other assistive technology. Compliance is validated using the PAC (PDF Accessibility Checker) tool.

What is the Matterhorn Protocol?

The Matterhorn Protocol is a PDF Association publication that defines 136 specific conformance checkpoints for PDF/UA compliance. It divides these into machine-verifiable checkpoints (checkable by automated tools) and human-verifiable checkpoints (requiring human review). It is the primary framework used by accessibility auditors and validation tools like PAC to assess PDF/UA conformance.

How do I check if my PDF is accessible?

Download and run the PAC (PDF Accessibility Checker) — free Windows tool that validates against all machine-verifiable Matterhorn Protocol checkpoints. For command-line batch checking, use veraPDF (open source). After automated validation, test manually with a screen reader — NVDA + Chrome is the recommended free combination.

What is the difference between a tagged PDF and a PDF/UA-conformant PDF?

A tagged PDF has a logical structure tree — a necessary but not sufficient condition for accessibility. A PDF/UA-conformant PDF is a tagged PDF that additionally satisfies all the specific requirements of ISO 14289: correct heading hierarchy, proper reading order, complete alt text, correctly structured tables, accessible forms, and no inaccessible embedded content. PDF/UA is the superset; a tagged PDF without full PDF/UA conformance may still have significant accessibility failures.

Try it in your browser

The PDF tools and standards discussed at this event are available free — no account needed.