Preparing Scanned Documents for OCR: A Practical Checklist

Before you run OCR

Use this checklist to maximize accuracy before opening the Basic OCR tool.

Scan quality

• [ ] Resolution is at least 300 DPI
• [ ] Text is sharp — not blurry when zoomed to 100%
• [ ] No significant shadows across the text area
• [ ] Pages are straight — not noticeably tilted

Document condition

• [ ] No torn or folded corners covering text
• [ ] No coffee stains or marks obscuring words
• [ ] Handwritten annotations are separated from typed text where possible

Image preparation

• [ ] Very large images (5+ MB) are compressed with Compress Image at quality 85%+
• [ ] Images are compiled into a PDF using Image to PDF

Language settings

• [ ] The correct document language is selected in the OCR tool
• [ ] For mixed-language documents, the dominant language is chosen

---

After OCR runs

Verify critical data

• [ ] Invoice numbers, totals, dates match originals
• [ ] Names and addresses look correct
• [ ] No obvious character swaps (O/0, l/1, rn/m)

Check structure

• [ ] Multi-column documents: verify reading order is logical
• [ ] Tables: confirm row/column alignment in output
• [ ] Headers and footers: verify they did not duplicate into body text

Handle failures

• [ ] Identify pages with poor output
• [ ] Rescan or manually correct those pages
• [ ] Do not re-run OCR on the full document for a few bad pages

---

What OCR cannot reliably handle

Even with perfect preparation, expect limitations with:

• Handwritten text
• Text printed over background images or watermarks
• Very small type (below 8–9pt equivalent)
• Decorative or script fonts
• Complex tables with merged cells

For these cases, manual transcription is more reliable than attempting to improve OCR output further.