How-to · Compress & OCR
How to make a scanned PDF searchable
By the PDFBetter teamPublished 6 min read
Short answer
To make a scanned PDF searchable, run OCR (optical character recognition) on it with the right language selected. OCR adds an invisible text layer behind each page image, so Ctrl+F (Cmd+F on a Mac) finds words and you can copy text, while the page looks exactly as it did. Then search for a few words you can see to confirm it worked.
Do it now, without leaving this page
Drop a file hereLet go to start
or paste with Ctrl/⌘+V
On this page (8)
Why can't you search a scanned PDF?
A scanner, a phone scanning app or a fax service doesn't save words. It saves a photograph of each page and wraps the photos in a PDF. To you the page shows an invoice number; to the computer it is a grid of grey and white pixels. Search, copy and paste have nothing to work with.
You can test any file in five seconds. Press Ctrl+F (Cmd+F on a Mac) and search for a word you can clearly see on the page. If nothing is found, or your cursor can't select a single word, the file needs OCR.
Some scanners and apps already add a text layer, so text selects. That layer is only as good as the software that made it. If search finds some words but not others, the existing layer has recognition errors, and running OCR again with the right language usually does better.
What does a searchable PDF actually contain?
A searchable scan has two layers. On top is the original page image, untouched, so it prints and looks exactly as before. Behind it, invisible, is the recognized text, placed where each word sits on the page. When you search, the reader looks in that hidden layer and highlights the matching spot on the image.
| You want to | Works after OCR | What to use instead or next |
|---|---|---|
| Find an invoice number with Ctrl+F | Yes | Nothing else needed |
| Copy a paragraph into an email | Yes, check it for misreads | Nothing else needed |
| Find the file later from your computer's search | Yes, in desktop search tools that index PDF text | Name files clearly too |
| Ask questions about the document with AI | Yes, AI tools read the text layer | Chat with PDF |
| Change the words on the page | No, the page is still an image | Edit scanned PDF |
| Get an editable Word file | Not directly | OCR, then PDF to Word |
Make a scanned PDF searchable, step by step
- 1
Turn pages the right way up
Recognition works best on upright, straight pages. Fix sideways or upside-down pages with [Rotate PDF](/tools/rotate-pdf/) first.
- 2
Clean up poor scans
If pages are tilted, grey or faint, run [Restore scanned PDF](/tools/restore-scanned-pdf/) first. Clean pages are recognized more reliably.
- 3
Upload to OCR PDF
Add a PDF, or a JPG, PNG or TIFF image. Images are placed on PDF pages before recognition.
- 4
Choose every language on the page
Accents and non-Latin scripts depend on this setting. A Spanish invoice with English line items needs both.
- 5
Run OCR
Each page counts toward your monthly OCR allowance: a few pages without an account, 20 on Free and 1,000 on Pro.
- 6
Test before you download
Search for two or three words you can see, including a number. If they are found, download the searchable PDF.
How do you make a whole folder of scans searchable?
The usual case isn't one file. It's a year of receipts for an expense report, a box of old contracts, or every page a client faxed over. A little preparation saves OCR pages and makes the results easier to find later.
- Combine what belongs together. Merge related scans into one file with Merge PDF, or turn phone photos into one PDF with JPG to PDF, then run OCR once.
- Drop pages that don't need text. Blank backs of double-sided scans and cover sheets still count as pages. Remove them with Remove pages before OCR.
- Name files for the search you'll run later. “2026-03 Acme invoice 4471.pdf” is found by date, vendor and number, even before anyone opens it.
- Keep one language per batch where you can, so each run has the right language setting.
| Document | Test search after OCR | Watch out for |
|---|---|---|
| Supplier invoices | Invoice number and total | Numbers in small print or inside stamps |
| Signed contracts | Party names and the effective date | Signatures and handwritten dates aren't recognized as text |
| Receipts | Merchant name and amount | Thermal paper fades; restore faint receipts first |
| Medical or school records | Name and record date | Forms with boxes can scramble reading order |
Which PDFs need OCR, and which don't?
Not every PDF that came from a scanner needs OCR, and not every PDF that needs it came from a scanner. Where a file came from tells you what to expect:
| Where the PDF came from | Needs OCR? | Why |
|---|---|---|
| Exported from Word, Google Docs or an accounting app | No | The text is already real text; search works |
| An office copier's scan-to-email | Usually | Most copiers save page images; some add a basic text layer |
| A phone scanning app | Often | Depends on the app; test with Ctrl+F |
| A fax service | Yes | Faxes arrive as page images |
| A PDF someone printed and scanned back in | Yes | Printing and scanning turns text into pixels |
| Photos of pages saved as a PDF | Yes | Every page is a photograph |
Mixed files are common too: a contract exported from Word with a scanned signature page at the end. Search works on the typed pages and fails on the scan. To spend OCR pages only where they're needed, pull the scanned pages out with Extract pages, run OCR on them, and merge them back in.
How do you check that OCR got it right?
Recognition errors are invisible. The page image still shows the right word, while the hidden text says something else, and you only notice when a search comes back empty. So check the hidden layer directly:
| Symptom | Likely cause | Fix |
|---|---|---|
| Search finds some words but not others | Misread words | Check the language setting, restore the scan and run OCR again, or use AI OCR |
| Accented letters come out wrong | Wrong or missing language | Select every language printed on the page |
| Copied text is jumbled | Columns read in the wrong order | Copy one column at a time, or convert with PDF to Word and fix the order |
| Nothing is found on one page | The page is upside down or sideways | Rotate it and run OCR again |
| Tables come out as one long line | Tables are recognized as lines of text | Use PDF to Excel when you need the cells |
Search for numbers, not just words
A reader can guess a misspelled word from context; a misread digit in an invoice total or an account number is silent. Always test at least one number you care about.
When is standard OCR not enough?
OCR PDF uses the Tesseract engine, which handles clean printed text well. Handwritten notes, faint faxes, dense tables of figures and pages with stamps over text are harder. For those pages, AI OCR runs the same recognition and then a vision-model correction pass. It costs AI credits, so use it only on the pages that need it.
If you need to change what the scan says rather than search it, open it in Edit scanned PDF. For a broader walkthrough of recognition settings, see our guide on how to OCR a PDF.
Try OCR PDF on your own file.
No account needed to start, and tools that run in your browser never upload your file. Keep going on the free plan, or pick a plan when you need more.
Questions people ask
Does making a PDF searchable change how it looks?
No. OCR adds an invisible text layer and leaves the page image untouched, so the file looks and prints exactly as before.
How many pages can I OCR for free?
A few pages without an account, 20 pages a month on the Free plan and 1,000 on Pro.
Can I make a photo of a document searchable?
Yes. Upload the JPG or PNG directly. Shoot it flat, in even light, with the page filling the frame for the best recognition.
Will the searchable PDF work in other readers?
Yes. The text layer is a standard part of the PDF, so search and copy work in any PDF reader.
Can I search a password-protected scan?
Unlock it first with Unlock PDF, run OCR, then protect it again if it needs to stay encrypted.
Do I need OCR before using AI tools on a scan?
Yes. Summaries, chat and AI search read the document's text, and a scan without OCR has none.
Sources
- PDFBetter plans and OCR page limits — pdfbetter.com, checked
- PDFBetter security and data retention — pdfbetter.com, checked