A scan is an image inside a container

When a sheet of paper goes through a scanner or a phone camera, the result is one large image per page, wrapped in a PDF so it prints and pages correctly. There is no character data in it: the letter A is a pattern of dark pixels, exactly like the letter B.

A PDF can hold both at once — an image and an invisible text layer placed over it, which is what OCR produces. Most scans have only the image. That is why a 20-page scan can be 30 MB while a 20-page typed document is 300 KB.

What conversion can and cannot do

Converting a PDF to Word or to text extracts the text layer. If there is no text layer, there is nothing to extract, and the honest outcomes are: a document containing the page images, or a failure.

A converter that "works" on a scan and produces a Word file with pictures in it has not read the page — it has copied it. That can still be useful if the goal is an editable layout, but it does not give you searchable text, and it is not what most people mean by "convert".

This site does not run OCR. A scan stays a scan here: the conversion copies what is there and says so instead of inventing words. If you need searchable text, use an OCR tool — and check whether it runs locally before you upload a document you care about.

How to tell what you have

Three checks settle it in under a minute:

  • Select a paragraph. If it highlights, there is a text layer and conversion will work.
  • Search for a word you can see on the page. No result means no text layer.
  • Look at the file size per page. Hundreds of kilobytes or more per page is a scan; tens of kilobytes is text.

If you do need OCR

OCR recognises characters from pixels and writes a text layer back into the PDF. It is a different job from conversion, it is slower, and its accuracy depends on the scan: 300 dpi, straight, high contrast, no handwriting. Expect to correct names, numbers and anything set in a light font.

If the scan is only needed for reading, a PDF-to-image conversion is often enough, and it is faster and lossless compared with OCR.