When English OCR on a PDF works
OCR is the optional box on PDF to Word, not a separate promise that any scan becomes a perfect document. Leave it off when the PDF already has real text. Turn it on only for pages that are pictures of paper, and only for English.
When you need it
If you can highlight words in a normal viewer, you do not need OCR. The text layer is already there, and the ordinary extract is more accurate than a guess from pixels.
If a scanned contract comes out empty, the page is a picture. Tick English OCR. The engine is Tesseract, loaded when you ask for it. It needs a network connection to download. The page is marked in the DOCX with [English OCR] so you can see which text was guessed.
What it misses
- Small type. Footnotes and fine print are often incomplete.
- Handwriting. This is a printed-English engine, not a handwriting reader.
- Any language other than English. Do not expect a usable result for a German, Spanish, or Chinese scan.
- The layout. OCR does not rebuild columns, tables, or the page design. You still get lines of text. PDF to Excel will not OCR a scanned table either.
The file:// block
OCR does not run when the browser opens the tool as a local file (file://). Chrome blocks the worker in that situation. Selectable text on a digital PDF still works. OCR works on the live site, where the page is served over https.
Double-clicking a saved copy of the HTML and expecting OCR is the usual way to hit this. Use the page on pctoolsonline.com. A local preview of the rest of the tool, including copying a real text layer, can still succeed. Only the OCR worker is blocked.
The same 80-page cap and the same password refusal apply. OCR does not unlock a protected file, and it does not read past the cap.
A sane order
Try the extract with OCR off. If the pages you care about are empty and they are printed English, turn OCR on and read the result as a draft. Correct names and numbers by hand. Do not file the OCR output as a certified copy of the original.
If you compress the scan first, you still have a picture, only a softer one. OCR can still run, but softer small type is harder to read. Prefer the original scan.
A mixed PDF is common: page 1 is a digital cover with real text, pages 2 and 3 are scanned attachments. Leave OCR on only if you accept that the digital page is still read from its text layer and the picture pages are guessed. The [English OCR] marker shows which pages were guessed. Check those pages against the scan. Account numbers and dates are where a wrong character matters.
The engine is English. A printed Spanish or German page will produce something, and that something is not a translation and not a reliable transcript. Do not turn the box on to 'see what happens' and then paste the output into an email as if it were the document.
Questions
Should I leave OCR on for every PDF?
No. Digital text is cleaner without it. Use OCR for picture-only English pages.
Why does OCR fail when I open the HTML from my desktop?
The worker is blocked on file://. Use the live site.
Will it read a handwritten note?
No. Handwriting, small type, and languages other than English are outside what this OCR is for.