How to make a scanned PDF searchable
To make a scanned PDF searchable, run it through OCR: it reads the text in the page image and lays an invisible, selectable copy over the scan. The page looks identical — English only, and words it is unsure of are dropped, not buried.
To make a scanned PDF searchable, run it through OCR: it reads the text in the page image and lays an invisible copy of that text over the scan. The page looks byte-for-byte identical, but you can now select, copy and search it. A scan is a photograph of a document, which is why your cursor finds nothing in it until OCR writes the characters back.
How to do it
Open the OCR tool
Go to OCR PDF. It runs in your browser — no account, no install, and your file is never uploaded.
Choose your scanned PDF
Pick the PDF you want to make searchable. The first run downloads the ~7 MB recognition engine from this site, which your browser then keeps cached; a password-protected PDF has to be unlocked first.
Click Make PDF searchable
Each page is rendered and read in turn — you will see "page 3 of 12" tick along. Recognition is the slow part, so you can leave the tab to work.
Save the result
The searchable PDF downloads on its own, with _searchable added to the name. The tool reports how many pages it read and the engine's own confidence score.
What "searchable" actually does to the file
It adds text; it does not change the picture. The recognised words are drawn over each page at zero opacity, so the scan renders exactly as before, and search and selection read those invisible characters instead of the pixels. Nothing is re-encoded, so the image cannot get worse in the process — the file just gains a few kilobytes of text.
Searchable is not editable. OCR does not turn the scan back into a document you can retype or reflow — the original page image is still there underneath, and the recognised words are anchored to where they sit on it. If you need to change the wording you are rebuilding the document, not OCRing it. For the full picture of what survives, see what OCR can and cannot recover.
Why are some words missing?
Anything the engine was not sure of is dropped, not hidden. Words it scores below 30% confidence — and any word its built-in font cannot write — are left out and counted for you, rather than buried invisibly where they would pollute every later search. So a page that comes back mostly blank is a fact the tool tells you, not a mystery to find later.
Does it work in other languages?
English, in Latin script, only. The recognition model is English, and the invisible layer is written with a standard WinAnsi font — the same wall that turns unsupported characters into question marks. Cyrillic, CJK and Arabic are not recognised, and any character the font cannot hold is dropped rather than mangled.
What if a page already has text?
Pages that already carry selectable text are skipped. Reading them again would stack a second, slightly-wrong copy of every word under your cursor, so a mixed PDF comes back with only its scanned pages changed. If every page already has text, the tool stops and tells you there is nothing to add.
How do I get the best result?
Scan quality sets the ceiling. Each page is read at about 216 DPI, and accuracy falls off fast below roughly 150 DPI, so a clean, straight, high-contrast scan beats a phone photo every time — small print, tight columns and handwriting come out worse. Check the confidence number the tool reports, and proofread anything you are going to rely on.
