Most "OCR a PDF" tutorials assume you have Acrobat Pro. You probably don't, and you don't need it.
This guide covers:
- What OCR actually does (and what it can't do).
- Free OCR for one-off scans on each major OS.
- When to pay for OCR (and when not to).
- How to verify the OCR worked before you ship the file.
What OCR actually does
OCR — Optical Character Recognition — takes a scanned image (a PDF page that's really a photograph of paper) and adds a hidden text layer underneath. The page still looks identical to the human eye, but search, copy-paste, and screen readers can now see the words.
What it can't do:
- Read handwriting well. Handwriting OCR is a separate, much harder problem. Even paid tools top out at ~70% accuracy on cursive.
- Recover text from a corrupt PDF. If the original document is broken, OCR won't reconstruct it.
- Reconstruct the original layout. OCR adds a text layer; it doesn't recover columns, tables, or exact positioning. For layout-critical PDFs, you need the original document or a PDF editor.
Free OCR, by OS
macOS — Preview
Preview (the built-in PDF reader) does OCR on every macOS since 10.14 Mojave.
- Open the scanned PDF in Preview.
- File → Export → choose PDF as the format.
- Click Show Details to expand the export options.
- Enable Enable text recognition (also labeled "Embed text").
- Save.
The output PDF has a search-and-copac text layer, ~5–15MB larger than the image-only original.
Windows — Microsoft Lens (free) or OneNote
- Microsoft Lens (iOS/Android, free) photographs paper, runs OCR, and exports to PDF or Word.
- OneNote (free with any Microsoft account): paste the scanned image into a page, right-click → Copy Text from Picture. Paste into Word and save as PDF.
Linux — Tesseract
Tesseract is the open-source OCR engine. The GUI wrappers vary by distribution; on Ubuntu, sudo apt install ocrmypdf installs the most popular wrapper.
``
ocrmypdf input.pdf output.pdf
``
Adds a text layer to the scanned PDF, with the option to compress images. Output is the same size or smaller than the input, with full text-searchability.
When to pay for OCR
The free options above cover the 80% case. You should pay for OCR when:
- You have bulk batches. A service like DocuScan or AWS Textract can process thousands of pages overnight.
- You need layout fidelity. Adobe Acrobat Pro, ABBYY FineReader, and ReadIris reconstruct tables and columns that free tools lose.
- You need legal-grade accuracy. Court filings, immigration paperwork, and medical records often require OCR confidence scores above 99% — only paid services routinely meet that bar.
How to verify the OCR worked
- Search test. Open the PDF and press Cmd/Ctrl+F. Search for a word you can see in the document. If it highlights, OCR worked.
- Copy-paste test. Try to select and copy a sentence from the document. If the selection snaps to a text layer (not the image), OCR worked.
- File size sanity check. A properly OCR'd 10-page scan should grow by 200KB–2MB (the text layer). If the file grew by 0 bytes, OCR didn't run.
The honest alternative: just keep the original
If the document you have is already a clean digital PDF (not a scan), OCR won't help — there's no image to recognize. The fastest "OCR" is to find the original and use it.
If you can't find the original, free OCR above will work. If you need it done in volume or with high accuracy, paid.
Privacy
Every OCR tool above runs OCR on your machine (macOS Preview, Tesseract, Microsoft Lens) — no upload required. Online OCR services require uploading; if you're OCRing a confidential document, prefer the self-hosted route.
DocShop doesn't OCR yet — but How to convert a PDF to Word covers the adjacent workflow for documents you have in PDF form and want in Word.