Documents · PRACTICAL GUIDE
Cannot copy text from a scanned PDF? Use OCR, then check the result
A scan often contains a photograph of a page rather than the text itself. OCR can make that content easier to search or copy, but its output is a draft to verify—not a trustworthy replacement for the source.
Start with the safe checks ↓
THE QUICK ANSWER
Here’s the short version.
Try selecting a line and searching for a visible word. If neither works, use a supported OCR workflow such as current Chrome’s local PDF text extraction. Check the copied result against the page, especially names, amounts and table columns.
LET’S DO THIS
Work through these checks
Check whether text already exists
Open the PDF in a trusted viewer and try selecting a sentence. Search for a clear printed word. If text is available, copy a short sample before assuming a new OCR pass is needed.
Try supported local extraction
Current Chrome documentation describes on-device OCR for scanned PDFs. Open a local copy in an up-to-date Chrome viewer and allow extraction time. Availability and handling may differ by document or browser version; verify that selection actually works.
Improve the source if needed
For unreadable results, return to the page image. Straighten the document, improve focus and avoid glare before scanning again when possible. An enlarged blurry image does not recover letters that were never captured clearly.
Review a small sample
Copy one paragraph into a plain text editor and compare it with the original. Check punctuation, line order and easily confused characters. For a table, check each value’s column rather than trusting the pasted layout.
Verify the saved result separately
If you need a searchable PDF rather than copied text, save through a tool that explicitly supports that output. Reopen the saved file in another viewer and search again. Selectable text in one open tab is not proof of a portable saved text layer.
Recognition is not understanding
OCR identifies characters from an image. It does not establish that a total is correct, that a field belongs to the right person or that a missing word can be safely guessed. A perfectly plausible output can still be wrong in the one place that matters.
Use a practical review order: identifiers first, dates and quantities second, surrounding prose third. For a receipt, compare the individual amounts with the image before using them elsewhere. For a letter, check names and reference numbers before sending a reply.
Keep the original beside the extracted text
Name your source and reviewed output distinctly. If a colleague questions a value later, you should be able to point back to the actual page. Do not silently replace a doubtful word with your best guess; mark it for review or revisit the scan.
For tables, a list of numbers is not yet a usable spreadsheet. Restore the row and column relationships deliberately. A simple review sheet with a page reference and a checked-status column is more useful than a confident but unverified bulk paste.
Choose the smallest workflow for the job
If the paper is still in front of you, Use My Phone Document Camera can capture it and extract locally recognised text for review. Its current OCR is English-trained, and PDF export is a Pro/Studio capability. Do not assume that export automatically produces a searchable text layer.
If you already have a good scanned PDF, recapturing it through a phone is unnecessary. Start with the file and a supported local OCR route. Before using any cloud extraction service, understand where the document goes and whether you are permitted to upload it.
THE THINGS PEOPLE ASK
Common questions
Why can I see words but not select them?
They may be pixels in a scanned page rather than stored text. An OCR workflow can recognise them, with errors that need review.
Does OCR guarantee a correct spreadsheet?
No. You must verify values and their row/column relationships against the source.
Is selectable browser text always saved into the PDF?
Do not assume so. Reopen the saved result elsewhere and test search if a portable searchable PDF is required.
Sources & further help
Check the documentation for your version. Menus and available options can change.
YOUR NEXT MOVE
Still stuck? Narrow it down.
Note the exact error, what changed before it started, and which checks worked. That makes the next support conversation much more useful.
