A scan is a photograph of a page, so there are no words in it for a converter to take — you get a Word file full of pictures, or an empty one. Run the PDF through OCR first so the text is recognized, then convert the searchable copy to Word and tidy the layout.
Runs in this tab, on Tesseract (WebAssembly), pdf.js and pdf-lib, so your file is read from your own device and never sent anywhere. Check it yourself
How do I fix it?
1Check whether there is textPress Ctrl+F (Cmd+F on a Mac) and search for a word you can see on the page. No match means the page is a picture.
2Turn any sideways pagesFix the orientation first; recognition on a sideways page is mostly guesswork.
3Run it through OCR PDFPick the language the document is written in. Every page is read in your browser and a text layer is laid invisibly over it — the pages look exactly as before, but now hold words.
4Spot-check the figuresSearch for a few names and numbers you can see. Clean printed text recognizes well; faint copies, stamps and handwriting less so.
5Convert the searchable copyNow PDF to Word has words to take. Expect to tidy the layout: OCR recognizes text, not styles, so headings and columns may need a pass.
Why does this happen?
It is a scan, not a document. Paper put through a scanner or photographed with a phone is stored as images. A converter looks for characters, finds none, and either gives up or pastes the pictures into Word.
Only some pages are scanned. A typed contract with a scanned signature page, or a report with a photocopied appendix, converts partly: the typed pages come out as text and the scanned ones as pictures.
An earlier OCR was poor. Some scanners add a text layer automatically. If it was badly recognized, the PDF looks searchable but converts to nonsense, because the converter takes the bad text layer as it is.
The pages are sideways or upside down. Recognition reads lines of text the right way up. A page scanned at 90 degrees reads poorly, or not at all, until it is turned.
Questions
Can I convert a scanned PDF to Word for free?
Yes, in two steps: make it searchable with OCR PDF, then convert the result with PDF to Word. Both run in your browser, with no upload or account.
Why is my Word document just pictures of the pages?
Because the PDF is a scan with no text layer, so the converter could only carry the images across. Add the text with OCR first.
Which languages can OCR read?
English, Spanish, Portuguese, French, German and Indonesian. Other scripts, such as Hindi, Arabic, Chinese and Cyrillic, aren’t supported yet.
Does OCR work on handwriting?
Poorly. It is made for printed text; handwritten notes come out with many errors and need checking word by word.