Make every scanned PDF searchable in Google Drive

The short answer

A scanned PDF is a photograph of a page, so search can't see the words in it. In Google Drive, right-click the scan and choose Open with Google Docs; Drive runs text recognition and creates a Doc containing the words, which Drive search can then find. Do that for the one document you need, prove the search works, then fix your naming so every new scan says what it is. For a big backlog, a free tool called OCRmyPDF adds an invisible text layer to the PDFs themselves, in bulk.

Why search can't see inside a scan

Years of paperwork gets scanned dutifully and dumped into Drive, where each file is a picture, not text. The scanner photographed the page and nothing ever read it. So when an accountant, a solicitor or a customer asks for a specific document, you know you scanned it. You just can't put your hand on it without half an hour of scrolling and squinting at thumbnails.

Google's help on Drive search is plain about what search actually matches: file names and text within the file. A picture of an invoice contains neither the supplier's name nor the invoice number as text, so search has nothing to work with beyond a name like Scan_20230414_093122.pdf. Google doesn't promise that search can see inside a picture, and you shouldn't rely on it to.

Optical character recognition (OCR) is the fix. Software reads the image and stores what it finds as real text, either in a companion document or as an invisible layer inside the PDF itself. The steps below use that one idea three ways: for the single document you need right now, as a filing habit, and against the pile.

1. Fix the one document you need today

Open Google Drive on a computer, find the scan, right-click it and choose Open with Google Docs. Drive runs its text recognition and opens a new Google Doc holding the words it managed to read. The original PDF stays exactly where it was; the Doc sits alongside it as the searchable copy.

Google's help sets out what works: PDFs or photo files (.jpeg, .png and .gif), 2 MB or smaller, text at least 10 pixels high, and the page the right way up (rotate it before uploading if it isn't). It detects the language by itself. Formatting mostly doesn't survive the trip; bold and italics usually do, while lists, tables and columns usually don't. That's fine, because you don't need the layout. You need the words.

Read the Doc through once. A clean office scan usually comes out close to perfect. A faded receipt won't, so correct anything you'll want to search for later, especially names and reference numbers.

2. Prove you can find it

Go back to Drive's search box and type a phrase you know is in the document, in quotes for an exact match. The Doc you just made should come straight back. A few operators make search sharper, and they can be combined. One thing to know: before: and after: match when the file was modified, not the date printed on the document, so a 2019 invoice scanned last month counts as last month.

If nothing comes back, the recognition probably failed quietly. Open the Doc and check it contains selectable text rather than just a picture of the page. If the text is gibberish, the scan quality is the problem, and there's more on that at the end.

"invoice 4471"
type:document "van repair"
after:2023-01-01 before:2024-01-01

3. Name files by what they are, not when they arrived

The scanner's default file name records the one thing you'll never search for: the moment of scanning. Rename the scan (and its Doc) so the name answers the question you'll actually ask later. Who it's from, what it is, and any reference that matters.

Keep folders broad and few. Invoices, Contracts, Insurance and Staff will beat a tree five levels deep, because once names and text carry the information, search does the narrowing. A file can only sit in one folder, but a good name can mention the supplier, the job and the reference all at once.

Scan_20230414_093122.pdf  →  Hendersons invoice 4471 van repair.pdf

4. Make it the habit for every new scan

Deal with each new scan at the point of filing. Name it properly the moment it lands, and convert the ones you'll need to find by their contents. Not everything deserves the Docs treatment; a well-named PDF is findable by name alone, so save the conversion for contracts, quotes and anything carrying reference numbers.

If your scanner emails PDFs to you, move them into Drive and name them there rather than letting them stack up in an inbox. The habit matters more than any tool. A folder of properly named scans beats the cleverest search over four thousand files named after the day they arrived.

5. Clear the backlog with OCRmyPDF

The Docs trick is one file at a time with a 2 MB ceiling, which makes it the wrong tool for a decade of paperwork. For the pile there's OCRmyPDF, a free, open-source command-line tool whose whole job is adding a recognised-text layer to scanned PDFs so they can be searched. The words sit invisibly behind the page image, inside the PDF itself. The file looks and prints exactly as before, but Ctrl+F works in any PDF reader, and the text travels with the file wherever it goes.

It does need someone comfortable with a command line. On a Mac it installs through Homebrew, and on Ubuntu or Debian (including the Windows Subsystem for Linux; a native Windows build exists too) through apt. The basic command takes an input file and writes a searchable output file, and one find command will walk a whole folder tree and OCR every PDF in place:

OCRmyPDF only overwrites a file when a run succeeds, but back the folder up before any in-place batch anyway. Two options earn their keep: --rotate-pages fixes jobs where some pages were scanned sideways, and -l sets the language, because it assumes English unless told otherwise. The documentation also covers watched folders, so a scanner can drop files in one end and searchable PDFs appear at the other. For a Drive backlog, the workflow is download a folder, process it, upload the results back.

brew install ocrmypdf          # Mac (Homebrew)
sudo apt install ocrmypdf      # Ubuntu, Debian, WSL

ocrmypdf scan.pdf scan-searchable.pdf
find . -name '*.pdf' -exec ocrmypdf '{}' '{}' \;

If your office runs on Microsoft, not Google

Microsoft 365 has an answer too, but it's an admin feature rather than a right-click. The OCR service in SharePoint extracts printed or handwritten text from images and from scanned or hybrid PDFs, and indexes what it finds so search can match it. It has to be enabled by whoever administers your Microsoft 365, it's scoped to SharePoint sites, and it carries pay-as-you-go billing rather than coming included in the licence.

If that's more plumbing than your business wants, OCRmyPDF sidesteps the question. A processed PDF carries its text layer inside the file, so it's searchable with Ctrl+F in any PDF reader wherever you keep it, OneDrive included.

Where this gets hard

OCR is only as good as the scan. Handwriting, fax-quality copies, faded thermal receipts and pages photographed at an angle defeat it far more often than anyone admits. Google's guidance asks for sharp images, even lighting and text at least 10 pixels high, and plenty of real-world paperwork fails that test. What comes out is noise you'd never think to search for. Those documents need a better source copy, a human eye, or acceptance that the file name is all you'll ever have.

The backlog is a judgement problem wearing a technical costume. OCRmyPDF will churn through four thousand files without complaint, but it won't name a single one of them, and searchable-but-meaninglessly-named is half a fix. Deciding that this blur is the Hendersons invoice and that one is the lease variation is human work. Doing it wrong is worse than not doing it at all, because a misfiled document that looks filed stops anyone ever searching for it again.

And the command line has its own edges: installing on Windows, PDFs that arrive password-protected or digitally signed, documents in more than one language, and in-place batch runs against files you can't afford to lose. None of it is exotic, but I'd never point a batch job at the only copy of anything.

What you end up with

• Type a supplier's name or an invoice number into Drive and the document comes back
• A searchable copy of any scan that matters, made with a right-click
• New scans named by what they are, findable without opening a single file
• Backlog PDFs carrying a real text layer, searchable wherever the file goes

Facts on this page last checked 14 July 2026, against support.google.com, ocrmypdf.readthedocs.io, learn.microsoft.com.

Related fixes

If the pile runs to thousands, the scans are too rough to read, or you'd rather someone else did the naming and checking, that's a job I take on for small businesses, and the first conversation costs nothing.