Automatic vs. Manual: PDF Accessibility Compared Honestly
If you regularly need to make PDFs accessible, you face a fundamental decision: manual with Adobe Acrobat Pro, semi-automatic with classic tools, or AI-assisted with modern conversion solutions? This article compares all three approaches honestly — with realistic ranges, clear limitations, and practical calculations. There is no “100% automation” magic bullet; but there are pragmatic solutions that make sense depending on the use case.
The three approaches at a glance
| Criterion | Manual (Adobe) | Semi-automatic | AI-assisted |
|---|---|---|---|
| Time per PDF (20 pages) | 8–12 hours | 2–4 hours | Few minutes processing + validation |
| Original layout | Risk of layout changes during tagging | Layout adjustments common | Original layout preserved (not reconstructed) |
| Manual rework | Fully manual | 30–50% needed | Validation recommended; special cases need rework |
| Scalability | Low | Medium | High |
| Cost per document | $650–950 | $160–320 | $5–35 (depending on complexity) |
| WCAG 2.1 AA | Achievable with effort | Achievable with rework | High automation, validation should follow |
1. Manual: Adobe Acrobat Pro
How it works
- Open PDF in Adobe Acrobat Pro
- Tools → Accessibility → Full Check
- Build tag tree manually (1–4 hours)
- Define reading order (1–2 hours)
- Write alt text for images (1–2 hours)
- Mark table headers (30 min – 1 hour)
- Validate and rework
Advantages
- Maximum control: Every tagging decision is manually controllable
- Established method: 20+ years of practice
- Complex special cases manageable (e.g., mathematical formulas, legal special texts)
Disadvantages
- Very time-consuming: 8–12 hours per standard document for proper implementation
- License costs: ~$240/year for Acrobat Pro
- Personnel costs: Trained specialists with hourly rates of $60–120
- Layout risks: Tagging interventions can affect visual presentation
- Inconsistency: Different editors → uneven quality without clear QA processes
When it makes sense
For individual highly complex documents (scientific publications with mathematical formulas, very old scans, legal special texts), when maximum control is required and time is a secondary concern.
2. Semi-automatic tools (axesPDF, CommonLook)
How it works
These tools handle part of the tagging work automatically but leave gaps that users fill manually:
- Automatic tagging of basic structure
- Suggestions for reading order
- Dialog for alt text (user writes manually)
- Manual validation at the end
Advantages
- Faster than purely manual: 2–4 hours instead of 8+
- Structural assistance: Automatic suggestions save tagging time
- Higher consistency than pure manual work
Disadvantages
- License costs: $550–1,650/year
- Layout adjustments: Structures are partly rebuilt
- 30–50% rework: Manual effort remains for full PDF/UA conformance
- Training needed: Tools are complex and require onboarding
3. Hybrid PDF approach with barrierefrAI PDF Pro
What is Hybrid PDF?
The Hybrid PDF approach entered the market in 2026 as a new method. It was developed by the agency barrierefrAI and is currently available exclusively via the company’s own tool, barrierefrAI PDF Pro — other providers still rely on classic tagging with layout reconstruction. The core idea: PDFs are organized in two layers.
- Visual layer: The unchanged original (as vector or high-resolution JPEG)
- Accessibility layer: An invisible, tagged text structure for screen readers
Sighted users see the original — screen readers use the structured layer. The decisive advantage over classic methods: the layout is preserved because it is not reconstructed from tags but kept as a visual image layer. This eliminates the biggest risk of classic tagging workflows — layout drift, where the end result deviates visually from the original.
How does the creation process work?
- Upload: Upload PDF in browser
- Complexity analysis: Detection of layout, multi-column, tables, images
- Routing: Direct for simple documents; wrapping procedure for complex ones
- Tag generation: Automatic creation of the accessibility layer
- AI image descriptions: Alt-text suggestions via vision model
- Validation: Automatic check against PDF/UA and WCAG 2.1 AA
- Download and review: Check result, refine alt text if needed
Advantages
- Original layout preserved: Since not reconstructed, no layout drift occurs
- Processing time: Few minutes for simple documents; complex ones take longer
- High automation level: Many documents can be converted with minimal manual effort
- Scalable: Batch processing of larger inventories possible
- Cost-efficient: Typically $5–35 per document, depending on complexity
- AI image descriptions: Automatic alt-text suggestions, manually refinable
Limitations (honestly)
- Slightly larger files (~10–20% more) due to additional layer
- Complex table semantics, mathematical notation, and chaotically structured PDFs often require manual rework
- Context-dependent alt text (e.g., “Mayor X opening the new building” instead of “person in suit”) usually needs to be sharpened
- No tool today can guarantee “100% automatic” and fully correct results — validation should always follow
Realistic calculation examples
Note upfront: The following figures are ranges based on industry experience values. Actual costs vary considerably with document complexity, hourly rates, and chosen tool.
Scenario 1: Mid-sized municipality with 200 PDFs/year
A medium municipality publishes about 200 PDF documents per year (press releases, applications, reports) averaging 12 pages with mixed complexity.
| Method | Time per PDF | Personnel costs ($75/h) | Tool license | Total cost/year (estimate) |
|---|---|---|---|---|
| Manual | 8–12 h | $120,000–180,000 | $240 | ~$120,000–180,000 |
| Semi-automatic | 2–4 h | $30,000–60,000 | $550–1,650 | ~$30,500–61,500 |
| AI-assisted + QA | 15–30 min | $3,750–7,500 | $550–1,650 | ~$4,250–9,000 |
For the AI-assisted variant, validation and any alt-text refinement is included in the 15–30 minutes. The ranges reflect different document complexity.
Scenario 2: Design agency with client projects
An agency produces about 30 client PDFs per month (annual reports, brochures, white papers) averaging 25 pages with medium to high design complexity.
| Method | Per PDF | Per month | Per year |
|---|---|---|---|
| Manual (Acrobat, $90/h) | 10–14 h × $90 | $27,000–37,800 | $324,000–453,600 |
| AI-assisted + 30 min QA ($90/h) | 0.5 h × $90 + $5–15 | $1,500–1,800 | $18,000–21,600 |
The difference is significant but assumes that 30 PDFs/month are actually processed and the designer’s hourly rate is applied. With lower volumes or simpler PDFs, the ratio shifts.
When is which method useful?
Manual with Adobe Acrobat Pro
Recommended when:
- You process highly specialized documents (complex mathematics, legal special texts, sheet music)
- You only have 1–3 documents per year
- You need maximum control over every tag detail
Semi-automatic tools
Recommended when:
- You’re already integrated into an Adobe workflow
- You have 5–20 documents per month and personnel for rework
- You need consistent structures across many documents but want detailed tag control
AI-assisted conversion
Recommended when:
- You have 5+ documents per month (ROI typically reached quickly)
- Original layout is important (marketing, design, tourism, municipal documents)
- You want to scale or process high PDF volumes
- You need a fast solution with traceable validation
What about quality?
A legitimate question: Does AI-assisted conversion really deliver comparable quality to manual processing? The answer is nuanced.
Where automated procedures excel
- Consistency: Every conversion follows the same rules
- Completeness: No “forgotten” image without alt text
- Scaling: 100 documents in the time it takes to manually do 1–2
Where manual processing has advantages
- Context-specific alt text: Describing a person not just as “man in suit” but as “Mayor Heinrich Müller at the 2024 opening”
- Special-case tagging: e.g., complex nesting in legal texts or mathematical MathML structures
- Tables with complex semantics: Multi-headers, nested columns, footer relationships
Both are not mutually exclusive: With AI-assisted preprocessing, the bulk can be automated, and alt text or special cases can be refined manually afterward — typically in Acrobat or directly in the tool.
FAQ
Does AI-assisted conversion work with InDesign exports?
Yes, well. InDesign exports often already have a basic structure that facilitates tagging.
What about scanned documents?
Before tag generation, OCR typically runs. Quality depends heavily on scan quality. Very old or poorly scanned documents may require manual rework.
Can I combine AI-assisted conversion with manual rework?
Yes, and that’s often the most pragmatic approach: AI does the structural pre-work, a human refines context-specific alt text and reviews special cases. Validation can be done in both steps.
How do I validate the result?
With free tools like PAC 2024 (PDF Accessibility Checker) or veraPDF (Open Source). Both check against PDF/UA. Adobe Acrobat Pro also has built-in validation tools.
Conclusion: Method choice depends on use case
There is no single best path — the right approach depends on volume, complexity, budget, and layout requirements. For many use cases with medium complexity and moderate to high volumes, the Hybrid PDF approach introduced by barrierefrAI in 2026 — available via barrierefrAI PDF Pro — is the most economical solution while also delivering the highest layout fidelity. For highly complex special cases (mathematics, legal special texts), manual processing remains relevant. Semi-automatic tools fill the middle position for team-integrated Adobe workflows.
Important: Even with the Hybrid PDF method, validation with independent tools (PAC 2024, veraPDF) should be standard practice.
Test it yourself:
No responses yet