PageFinderv4.2.11.22Sign in

Blog

How to search inside scanned PDFs in OneDrive

Why OneDrive finds some PDFs by their contents and not others, three ways to make scans searchable, and what to do when there are thousands of them.

November 19, 2026 · 6 min read · PageFinder team

PAIDSaved from Wordtext layer · search reads every wordOut of the scannera picture · search sees the file name

If you have ever typed a client's name into OneDrive search, watched it find the Word file and the email, and miss the signed contract you know is there, you have met the difference between a PDF with text and a PDF that is only a picture.

Why search finds some PDFs and not others

A PDF can carry its text as text, or it can carry an image of a page. A file saved from Word or exported from an accounting system has a text layer: every word is in the file as a word, and OneDrive's search indexes it. A file that came out of a scanner, a copier or a phone camera is an image. There is no text layer unless something has added one, and search cannot look inside a picture. The file name is all it has.

You can check any PDF in a few seconds: open it and try to select a line of text with the mouse. If the cursor selects words, there is a text layer. If it draws a box around the page, it is a scan.

the words selectonly a box
The select-a-line test: words select in a text PDF, a scan only gives you a box

Most archives are a mix. Everything born in the last few years is probably text; everything that was signed, stamped, faxed or scanned to file is probably not — and those are often the documents that matter most.

Three ways to make a scan searchable

1. Let Microsoft 365 do it, if your plan allows. Microsoft offers an optical character recognition service for OneDrive and SharePoint that an administrator turns on, billed per page processed, as part of its content services. Once enabled for a site or library, newly uploaded images and image-only PDFs get a text layer that search can use. Details, names and pricing change, so check your tenant's admin center and current plan before relying on it; on a personal OneDrive this option does not exist.

2. Add the text layer yourself. Adobe Acrobat, and a number of free tools, can "recognize text" in a scanned PDF and save it with a searchable layer behind the image. The page still looks like a scan; the words are now there for search to find. This works well for a few dozen files. It does not work for a few thousand, because someone has to open and re-save each one, and the re-saved file is a new version that other people's links may not point to.

3. Put the information in the name. The oldest trick: rename the file so that the client, the date and the kind are in the file name, and keep a consistent pattern. Search then finds the file without reading it. The cost is discipline, forever, from everyone, and it says nothing about what is inside the file beyond what was typed into its name.

What changes when there are thousands

Each method above is fine for the next scan. None of them is fine for the archive you already have: ten years of signed agreements, invoices and letters that nobody is going to open one by one and re-save. The honest answer for an archive is a tool that reads the scans where they are, from the page image, and does not require the files to be changed or moved.

That is what PageFinder does. It connects to OneDrive read-only, reads every file in the folders you choose — scans from the page image, born-digital files from their text — and catalogues each one: what kind of document it is, the date it is about, the companies and people it names, a short summary. Then you can ask across all of it ("which agreements with Northwind mention a termination fee?") and get the answer with the page cited, so you open the original and check. Nothing is copied or moved. You can see what it makes of one folder in about an hour: PageFinder for OneDrive.

A short checklist

  • Select a line of text in a suspect PDF. No selection means no text layer.
  • For new scans, turn on OCR where your plan offers it, or recognise text before filing.
  • Keep file names consistent; it costs nothing and helps every method.
  • For the archive, use something that reads the page image in place rather than a project to re-save ten years of files.