Convert a scanned PDF to editable Word

How to tell whether your PDF is scanned

Open the file in any viewer and try to select a line of text with your cursor. If a selection highlight appears over the words, the text is real and you can convert directly. If your cursor draws a rectangle across the page instead, or selects the entire page as one object, it is a scan.

A second check: use your viewer’s search function and look for a word you can plainly see. No result means no text layer.

Getting the best OCR result

Recognition quality is decided before you start, by the scan itself. Around 300 DPI is the sweet spot — below 200 DPI characters begin to merge, and above 400 DPI you gain little but a much slower run. Straight pages matter more than people expect; a few degrees of skew from a hand-fed scanner measurably increases errors, so deskew or re-scan crooked pages if you can.

Contrast helps too. A grey photocopy of a photocopy gives the engine less to work with than a clean black-on-white original. If your source is faint, scanning again in black and white mode often beats any amount of post-processing.

Set the language correctly. An English-language engine reading a German document will guess plausible English words at every accented character, and the result is harder to fix than the original.

What the Word file will look like

Expect the words, not the design. The export carries text and basic structure across; it does not attempt to be a pixel-perfect reproduction of the original layout. Multi-column pages, tables with merged cells, sidebars and footnotes are where it simplifies most, because reconstructing those from a flat page is genuinely hard and guessing wrong is worse than not guessing.

For a letter, a report or a contract, the output is normally something you can edit immediately. For a densely formatted form or a magazine page, treat it as a way to recover the text rather than the document.

Limits worth knowing

• OCR runs in your browser, which is slower than a server and much slower on a phone. A long document is worth starting on a laptop and leaving alone.
• Handwriting is not supported in any reliable way. The engine is trained on printed type.
• Accuracy on a good 300 DPI scan of clean printed text is high but never perfect — always proofread figures and identifiers.
• The Word export is best-effort on layout. Complex tables and multi-column designs will simplify.
• A password-protected PDF has to be unlocked before either step will read it.