By PDFKits Team — Published February 19, 2026
TL;DR: PDF accessibility means a screen reader can announce content in the right order: headings tagged as headings, images with alt text, defined reading order, and keyboard-operable forms. The legal floor is Section 508 compliance for U.S. federal work and WCAG 2.1 AA under DOJ rules, with WCAG 2.2 as the current W3C standard; the EU's European Accessibility Act took effect June 2025. Four levers — tags, alt text, reading order, forms — cover roughly 90% of compliance.
Three frameworks govern accessible pdf work in practice. The W3C's WCAG guidelines are the technical baseline — WCAG 2.1 AA is the level cited by the DOJ's ADA Title II rule (finalized April 2024) and by most state laws, while WCAG 2.2, the current recommendation, adds nine success criteria you should target for new documents. Section 508 binds U.S. federal agencies and their contractors and incorporates WCAG by reference. In Europe, EN 301 549 and the European Accessibility Act (EU 2019/882, in force since June 28, 2025) extend the obligation to private-sector digital documents.
All three regimes converge on PDF/UA (ISO 14289-1), the PDF Association-backed standard that operationalizes WCAG for the format: proper tag structure, alternate text on every meaningful image, defined reading order, and language declarations. A document that looks perfect on screen can still be a flat, unnavigable word-stream for the roughly 285 million people worldwide with visual impairments — and an untagged PDF is exactly that.
Best for: proposal teams, universities, HR departments, insurers, and any organization that publishes documents the public must be able to read.
Tags are the document's structural skeleton — an internal tree of H1, H2, P, Figure, List, and Table elements, analogous to semantic HTML. Without it, a screen reader gets a flat word stream: no heading navigation, no table semantics. Alt text makes images meaningful; the most common failure is a logo announced on every page header, pure noise. Reading order is defined separately from visual layout — a two-column page must read down one column then the next, and footnotes must not interrupt body text. Forms need programmatically associated labels, logical tab order, and required-field indications a screen reader can announce; a field "labeled" by a nearby asterisk is invisible to assistive tech. Master these four and you have covered the bulk of WCAG 2.2 pdf failures auditors actually flag.
Two testing habits separate teams that pass audits from teams that fail them. First, navigate the document by headings only (H key in NVDA, VO+Cmd+H in VoiceOver) — if you cannot reach every section that way, neither can your users. Second, tab through every form without touching the mouse: the focus order should match the visual order, and each field should announce its label and whether it is required. Five minutes of keyboard-only testing finds the defects automated checkers describe abstractly.
| Capability | PDFKits | Adobe Acrobat Pro | Smallpdf / iLovePDF | PAC 2024 |
|---|---|---|---|---|
| Cost | Free | $14.99/month | $9/mo / $48/yr | Free |
| OCR layer for scanned PDFs | Yes, in-browser | Yes | Yes (paid, cloud) | No — validator only |
| Edit the tag tree | No | Yes (granular) | No | No |
| PDF/UA validation | No | Limited checker | No | Yes — gold standard |
| Files stay on your device | Yes | Yes (desktop) | No — cloud upload | Yes |
| Page-level restructuring | Yes | Yes | Basic | No |
The honest division of labor: source documents and Acrobat Pro own tag editing; PAC owns validation; PDFKits owns the preparation and sanitization steps — OCR, page reordering, splitting, metadata cleanup — without uploading student records or HR files to a cloud service the way Smallpdf-style tools require.
Trusting the visual check. Tagging defects are invisible on screen. Always validate with a checker and a screen reader.
Color-only signaling. "Red = urgent, green = done" fails WCAG Success Criterion 1.4.1. Pair color with labels or icons.
ASCII tables. Columns built with spaces collapse into gibberish for screen readers. Use real table objects in the source.
Skipped heading levels. H1 → H3 breaks heading navigation. Keep a strict hierarchy.
Flattening forms that must stay navigable. Flattened fields read as plain paragraphs. If users need field-level navigation, keep the form interactive with proper labels — and check signing workflows, since some flatten silently.
Believing OCR equals accessibility. OCR makes text extractable; the document still needs tags, reading order, alt text, and a language declaration. OCR is step one of four.
No. Selectable text is necessary but not sufficient. Without tags, a screen reader cannot navigate by heading, skip decorative content, or announce table structure. Tagging is the qualitative difference between readable and usable.
WCAG 2.2 (October 2023) adds nine success criteria, mostly around focus visibility, input modalities, and authentication. Legal mandates still largely cite 2.1 AA, but new documents authored to 2.2 satisfy both — the criteria are backward-compatible.
Since the 2018 refresh, Section 508 incorporates WCAG 2.0/2.1 AA: tagged structure, alt text, reading order, language declaration, and accessible forms. Federal agencies and contractors must meet it for all public-facing and many internal documents.
Not at tag-tree level. PDFKits handles OCR, page reordering, splitting, and metadata cleanup in the browser; granular tag editing requires Adobe Acrobat Pro or re-export from a properly prepared source document.
PDF/A is the archival standard (long-term preservation, embedded fonts); PDF/UA is the accessibility standard (screen-reader compatibility). A document can be both — PDF/A-2u plus PDF/UA is the gold standard for government archives.
No — it is the mandatory first step. After OCR, the document still needs tags, reading order, alt text on figures, and a language declaration before a screen reader can use it properly.
Use at least two: NVDA on Windows (free, the de facto testing standard) and VoiceOver on macOS (built-in). If you can only run one, NVDA covers the largest real-user base.
Only modestly — typically 5–15% larger. The tag tree adds structure objects, not images or fonts, so the size cost is trivial compared to the usability gain.
Since June 2025 it covers products and services in scope — e-commerce, banking, e-books, transport — including the digital documents they ship. B2C businesses serving EU customers should treat customer-facing PDFs as in scope.
State the takeaway, not the geometry: "Q4 revenue exceeded Q3 by 12%, driven by North America" rather than "bar chart with eight bars." For data-dense charts, add a real data table in the document body as well.
OCR PDF — add a text layer to scans before tagging. Rearrange Pages — fix page-level order. Edit PDF — adjust visible content alongside Acrobat tag work. Clean Metadata — set title and language metadata before publishing. Extract Pages — isolate sections for targeted fixes. Split PDF — deliver per-chapter accessible files.