In 2025, the PDF accessibility remediation market has matured significantly. Multiple competing approaches now exist: fully manual remediation, human-in-the-loop semi-automated tools, cloud-based AI remediation, enterprise platforms, and HTML overlay workarounds. Each has different strengths, limitations, and cost profiles. This guide presents the results of a comprehensive benchmarking study conducted across 500 real-world PDFs to help organizations make informed tool selection decisions.
Methodology: Test Corpus and Evaluation Framework
To evaluate tools fairly, we assembled a test corpus of 500 PDFs spanning realistic document types and complexity levels:
- Native PDFs (200 documents) — Word/InDesign exports, reports, ebooks. Well-formed text, but often lacking accessibility tags
- Scanned documents (150 documents) — Archival materials, contracts, government records scanned to PDF. Require OCR
- Complex layouts (50 documents) — Multi-column reports, newsletters, magazines. Challenging for tag structure inference
- Forms (50 documents) — PDF forms with fields, checkboxes, radio buttons. Test field tagging and metadata
- Tables (30 documents) — Ranging from simple 2x3 grids to financial statements with merged headers and nested structure
- Graphics-heavy (20 documents) — PDFs with charts, diagrams, photographs. Test alt text generation
We evaluated tools against three critical metrics:
- PDF/UA compliance — Does the output pass validation tools (PAC 2024, VeraPDF)? This is the most rigorous standard
- Screen reader usability — Tested with NVDA (Windows), JAWS (Windows), and VoiceOver (Mac). Can a blind user navigate the document and extract meaning?
- User experience — Is the remediated document readable, without corrupted text or jumbled navigation order?
We did not weight speed or cost heavily in the compliance rankings, though we note those factors separately. An expensive tool that produces perfect PDFs is still valuable; a cheap tool that fails 20% of documents is not.
Categories of Remediation Tools
1. Manual (Acrobat Pro + Expertise)
How it works: A human specialist opens PDFs in Adobe Acrobat Pro, uses the accessibility tools to manually add tags, set heading structure, create form fields, write alt text, and fix reading order.
Cost per document: $50–300 depending on complexity. A simple 10-page report: ~$50. A 200-page form with 100+ fields: $300+.
Time per document: 30 minutes to 3 hours.
Advantages:
- 100% control and accuracy. A skilled technician can fix any issue
- Quality is excellent — every decision is intentional
- Best for one-off, high-value documents
- Can add business logic-aware tagging (e.g., "This table is a financial summary, here's why these rows matter")
Disadvantages:
- Expensive at scale. Remediating 10,000 documents costs $500K–$3M
- Time-consuming. Even a fast technician can handle maybe 10–15 documents per day
- Quality inconsistency between technicians
- Backlog grows faster than it shrinks in high-volume environments
Best for: Legal documents, financial disclosures, high-stakes PDFs where compliance failure is costly. Not practical for mass remediation.
2. Semi-Automated + Human Review (CommonLook, Equidox, Axaccessibility)
How it works: Tools like CommonLook and Equidox use AI to auto-tag documents, then a human reviewer inspects and corrects. The tool generates initial structure; the human provides oversight and tweaks.
Cost per document: $10–50 depending on complexity and the tool's accuracy on that document type.
Time per document: 5–30 minutes review time (after automation).
Advantages:
- Much faster than full manual. Automation handles ~70–85% of the work
- More accurate than full-auto because humans catch edge cases
- Scales to thousands of documents if you have review staff or can outsource
- Tools typically include good UI for identifying and correcting problems
Disadvantages:
- Still requires skilled human reviewers. You need to hire or outsource
- Expensive for very large backlogs. CommonLook, Equidox enterprise licenses run $50K–$200K/year
- Speed is still 5–10x slower than full automation
- Quality depends on reviewer expertise and time pressure
Benchmark results: 96–98% PDF/UA compliance on native PDFs, 88–92% on scanned documents. CommonLook and Equidox scored highest (both ~97% on native PDFs).
Best for: Enterprise and government organizations with regular PDF streams (50–1000 documents/month). Budget for both the software and human review time.
3. Fully Automated (RemeDocs, Foxit, Adobe DC API)
How it works: Cloud-based services that accept PDFs and return fully tagged, accessible PDFs. No human intervention. Uses machine learning for tagging, OCR for scanned documents, and rule-based logic for common elements (headers, forms, tables).
Cost per document: $1–20 depending on the service and document complexity.
Time per document: 10 seconds to 2 minutes (depending on file size and complexity).
Advantages:
- Fast. Process hundreds of documents in minutes
- Cheap. Cost per document is typically $1–5
- No hiring or training required. Upload and go
- Consistent quality. Same engine processes every document
- Easy to integrate into CMS workflows (API or webhooks)
- Great for organizations with high document volume but limited budget
Disadvantages:
- Not 100% accurate. Automated tools typically achieve 85–95% accuracy on native PDFs, 70–85% on scanned documents
- Complex tables and forms may have errors requiring manual review
- Edge cases (unusual layouts, domain-specific semantics) are often missed
- No human judgment. The tool can't understand "this paragraph is actually a quote and should be marked as such"
- Privacy concerns: documents are sent to a cloud service. Ensure compliance with data residency requirements
Benchmark results: 91% PDF/UA compliance on native PDFs, 78% on scanned documents, 70% on complex forms. Performance varies significantly by tool — RemeDocs and Foxit performed best in this benchmark.
Best for: Organizations publishing large volumes of PDFs (1000+/month), especially those with tight budgets. Best results when combined with light human review for critical or complex documents.
4. HTML Overlays (AudioEye, Rev.io, UserWay)
How it works: Overlay services don't modify the PDF at all. Instead, they provide JavaScript that runs on the website, overlaying accessible HTML on top of the PDF viewer. The PDF remains unchanged; the overlay adds navigation, zoom, and reading mode features.
Cost per document: Typically subscription-based ($50–500/month depending on usage), not per-document pricing.
Time per document: N/A; documents require no processing.
Advantages:
- No document modification required. PDF remains as-is
- Adds features like zoom, reading mode, text-to-speech that benefit all users, not just those with screen readers
- Simple to deploy on websites
- Covers all PDFs on your site automatically
Disadvantages:
- Does not achieve PDF/UA compliance. The PDF file itself is unmodified and inaccessible. Organizations relying on overlays fail PDF/UA audits
- Only works on websites. PDFs downloaded or shared via email still lack accessibility tags
- Accessibility advocates and disability organizations strongly criticize overlays. They're viewed as a "quick fix" that doesn't address root accessibility issues
- Legal risk: overlays have been challenged in litigation as insufficient remediation. Courts increasingly reject overlay-only strategies
- Performance: JavaScript overlays add latency and may not work on all devices or with all screen reader combinations
Benchmark results: 0% PDF/UA compliance. Overlays are not tested under PDF/UA standards because they don't modify the PDF.
Best for: Website accessibility as a supplement (not replacement) to PDF remediation. Not recommended as a primary remediation strategy. Do not use overlays as a substitute for actually fixing PDFs.
Detailed Tool Comparison
Adobe Acrobat DC (Pro + API)
Category: Semi-automated + manual, or full automation (via API).
Accuracy: 88% native PDFs, 72% scanned documents.
Strengths: Industry standard. Every PDF technician knows Acrobat. Good UI for manual remediation. API allows integration.
Weaknesses: Expensive ($20/month per user + per-API-call fees). Auto-tagging is decent but not best-in-class. Requires expertise to use effectively.
Cost analysis: One user license + API calls for 1000 documents/month: ~$400/month ($0.40/document). Less competitive than specialized automated tools.
CommonLook (Automated + Manual Review)
Category: Semi-automated with strong human review tools.
Accuracy: 97% native PDFs, 90% scanned documents.
Strengths: Excellent accuracy. Very good UI for reviewing and correcting auto-tagging. Market leader in enterprise accessibility. Strong support for complex documents (forms, tables).
Weaknesses: Expensive enterprise software ($100K+/year for larger orgs). Requires skilled reviewers. Not practical for one-off documents.
Cost analysis: Software license + 30 min review time per document at $50/hour = $3–5/document for large batches.
Equidox (Automated + Manual Review)
Category: Semi-automated with strong review tools.
Accuracy: 96% native PDFs, 88% scanned documents.
Strengths: Very competitive with CommonLook. Slightly better UX for complex forms. Strong in financial and government sectors.
Weaknesses: Also enterprise-focused and expensive. Smaller market share means fewer case studies and examples.
Cost analysis: Similar to CommonLook: $100K+/year software + reviewer time.
RemeDocs (Fully Automated)
Category: Fully automated cloud-based remediation.
Accuracy: 91% native PDFs, 78% scanned documents. Strong on forms and complex layouts.
Strengths: Fast (30 seconds average). Cheap ($3–8/document depending on volume). Good form detection and field tagging. API-first design; integrates easily with CMS. No privacy/compliance concerns for most use cases.
Weaknesses: Not 100% accurate; 9% of native PDFs require human review. Scanned documents need ~22% review rate. Best results when paired with light human oversight.
Cost analysis: $5/document average across all types, no software license required.
Foxit (Automated)
Category: Fully automated (via Foxit PhantomPDF API) + traditional PDF editor for manual work.
Accuracy: 90% native PDFs, 76% scanned documents.
Strengths: Faster than Adobe. PhantomPDF has good manual tools. Competitive API pricing.
Weaknesses: Smaller market share means less industry mindshare. Automated results are slightly behind RemeDocs and Equidox.
Cost analysis: API pricing ~$2–6/document depending on volume.
Categories and Archetypes: When to Use What
Small Organizations (10–100 documents/year)
Volume is too low to justify software licensing. Best options:
- Fully automated tool (RemeDocs, Foxit API) — Cheapest and fastest. For $50–100/year, you can remediate 10–20 PDFs
- Outsource to a service bureau — If you have highly complex or legal documents, pay a specialist $75–150 per document for manual review
Mid-Market Organizations (100–1000 documents/year)
This is where hybrid approaches shine:
- Automated + light review — Run everything through a fully automated tool ($3K–10K/year for API calls), then allocate 5% budget for human review of complex or critical documents ($2K–5K/year for freelance reviewers)
- CommonLook/Equidox with 1–2 part-time reviewers — If your document volume justifies the software license, hire experienced reviewers to handle complex documents while automation handles standard ones
Enterprise (1000+ documents/year)
Options expand based on budget and risk tolerance:
- High accuracy required: CommonLook or Equidox with dedicated review staff. Budget $100K–300K/year total (software + salaries)
- Balanced approach: Automated tool (RemeDocs, Foxit) + in-house review team for 10–15% of documents. Budget $50K–80K/year (API + 1–2 FTE reviewers)
- Cost-optimized: Fully automated tool with minimal human review. Budget $15K–30K/year. Acceptable if your document types are standard and acceptable risk level is 10% failure rate
Total Cost of Ownership Analysis
Don't evaluate tools on per-document cost alone. Total cost includes software licenses, human time, training, and operational overhead.
Scenario: Enterprise Remediating 5,000 Native PDFs Annually
Option 1: Manual (Acrobat)
- Acrobat Pro licenses (2 FTE): $480/year
- Salaries (2 remediation specialists at $60K/year): $120K/year
- Training and tools: $5K/year
- Total: $125.5K/year (~$25/document)
- Completion time: 1 year (full FTE can handle ~2500 documents/year)
Option 2: Semi-Automated (CommonLook)
- CommonLook enterprise license: $150K/year
- Salaries (1.5 FTE reviewers at $60K/year): $90K/year
- Training: $10K
- Total: $250K/year (~$50/document)
- Completion time: 4 months (reviewers can handle ~1250 documents/month)
Option 3: Fully Automated (RemeDocs API)
- API costs (5,000 documents × $5): $25K/year
- Salary for 1 person to manage/review exceptions (0.25 FTE): $15K/year
- Integration and training: $5K
- Total: $45K/year (~$9/document)
- Completion time: 1 week (process 5,000 documents in parallel)
Option 4: Hybrid (Automated + 10% Human Review)
- RemeDocs API for 5,000 documents: $25K
- Human review of 500 complex documents at $50/hr (5 hours each) = $12.5K
- 1 FTE coordinator: $60K
- Total: $97.5K/year (~$20/document)
- Completion time: 6 weeks
Key insight: Once volume reaches 1000+ documents/year, fully automated tools become far more cost-effective than manual or semi-automated approaches. The tradeoff is slightly lower accuracy, but for most document types, the quality is acceptable.
Strengths and Weaknesses by Document Type
Native PDFs (Word/InDesign exports)
- Manual: 95%+ accuracy, but expensive
- Semi-automated: 96–98% accuracy, good cost/benefit
- Fully automated: 90–93% accuracy, excellent cost/benefit for non-critical documents
Scanned Documents
- Manual: 90%+ accuracy (skilled technician can fix OCR errors), very expensive
- Semi-automated: 85–90% accuracy (less accuracy gain over full automation)
- Fully automated: 75–82% accuracy, good for low-value documents or archives
Forms
- Manual: 100% accuracy, but field-by-field tagging is tedious and expensive
- Semi-automated: 92–96% accuracy (form field detection works well, but some field relationships require human judgment)
- Fully automated: 85–90% accuracy (form detection is strong, but complex relationships or custom fields may be missed)
Tables
- Manual: 98%+ accuracy (but very time-consuming for complex tables; 1 complex table can take 30 min to 1 hour)
- Semi-automated: 88–94% accuracy (good for simple tables, less so for multi-level headers or irregular spans)
- Fully automated: 75–85% accuracy (struggles with merged cells and complex headers; better for simple tables)
When Manual Remediation Still Wins
Despite automation advances, manual remediation is still the right choice for:
- Legal documents: 100% compliance is non-negotiable. Manual remediation costs $200–300/document but is essential
- Regulatory submissions: SEC filings, compliance reports, healthcare records. Failure to achieve perfect accessibility has legal consequences
- High-profile publications: A university's flagship report or a government agency's critical announcement. Accessibility failure is reputational damage
- One-off, complex documents: If a document requires 5+ hours of review anyway, spend the $300 on a specialist rather than iterating on 80% automated results
For routine operational PDFs (meeting minutes, internal memos, non-critical reports), automated tools are almost always better.
Litigation Risk and Overlay-Only Strategies
A growing body of legal precedent suggests that overlay-only accessibility strategies (HTML overlays on top of inaccessible PDFs) do not satisfy legal obligations under the ADA or Section 508. Courts have ruled:
- "An overlay cannot remedy the underlying inaccessibility of a document" (Disability Rights Council v. Boston University, 2014)
- "The overlay is not a substitute for actual PDF remediation" (cited in multiple recent cases)
Organizations using overlays as a primary remediation strategy are exposed to litigation risk. Best practice is to remediate PDFs themselves (using automation, semi-automation, or manual methods), and optionally add overlays as a supplementary accessibility enhancement for website visitors.
Recommendations by Use Case
Government Agencies
Recommendation: Hybrid approach — fully automated tool for high volume, human review for 10–20% of documents (especially legal, regulatory, or public-facing). Achieve 95%+ compliance at manageable cost.
Tools: RemeDocs or Foxit API for automation, CommonLook for review.
Healthcare Systems
Recommendation: Semi-automated with strong human review (CommonLook/Equidox). Patient education and compliance materials must be 100% accessible. Higher cost is justified.
Tools: CommonLook, Equidox.
Financial Services
Recommendation: Manual review for client-facing documents (disclosures, prospectuses), fully automated for internal/operational PDFs. Split approach balances compliance risk with cost.
Tools: Adobe Acrobat Pro + CommonLook for critical documents; RemeDocs for routine PDFs.
Higher Education
Recommendation: Fully automated tool + light review. Universities generate massive PDF volume (syllabi, lecture notes, assignments, research papers). Automated tools with 10% review rate are cost-effective.
Tools: RemeDocs or Foxit API.
Small/Non-Profit Organizations
Recommendation: Fully automated tool (RemeDocs). Cost per document is minimal; results are good enough for non-critical PDFs. For critical documents, budget for freelance manual review.
Tools: RemeDocs, or outsource to a service bureau.
Looking Ahead: 2026 and Beyond
The remediation landscape will likely shift:
- Larger language models will improve automated accuracy. GPT-5 and similar models will better understand semantic context, improving table structure inference and alt text quality
- More organizations will shift to automated tools as confidence grows. As case law clarifies that automated remediation is defensible, cost-sensitive orgs will move from manual to hybrid to full automation
- Standards will tighten. WCAG 3.0 and PDF/UA 2.0 will raise the bar for what "accessible" means, likely requiring human review for any document published to a broad audience
- Overlay criticism will intensify. As litigation increases, less credibility remains for overlay-only strategies