How to OCR a PDF and Make Scans Searchable for Free

What OCR can and can't do

Good OCR converts clear, upright, high-contrast scans into highly accurate text. It struggles with handwriting, skewed photos, low resolution, and degraded print. Modern OCR reports a confidence score per page — treat low-confidence pages as needing manual review rather than trusting them blindly.

OCR recognizes characters; it does not reconstruct layout perfectly. Reading order, columns, and tables may need cleanup after recognition.

Two output modes: text or searchable PDF

Text extraction gives you the recognized characters as an editable, copyable file — right choice when you need the content, not the document.

A searchable PDF keeps the original scanned pages visually identical but embeds an invisible text layer behind them, aligned to where the words appear. The result looks exactly like your scan, but selection, search, and copy all work. Choose this when the document's appearance is part of its meaning — signed forms, stamps, letterhead.

Getting the best recognition quality

Small inputs dramatically change OCR accuracy:

• Scan at 300 DPI or higher — resolution is the #1 accuracy factor
• Keep pages upright; rotate before OCR if needed
• Flatten curved pages and avoid shadows across text
• Pick the right recognition language — it changes the model used
• Review confidence scores and re-check anything under ~80%

Privacy and OCR

Cloud OCR services receive the complete content of your document — every word, signature, and number. Browser-local OCR downloads a recognition model once and runs recognition on your device. For documents containing other people's information, the difference is not cosmetic.