How to Make a Scanned PDF Searchable with OCR

A scanned PDF looks like a document but behaves like a photograph. Ctrl+F finds nothing, you cannot copy a sentence, and converters produce empty files. The missing piece is a text layer.

OCR PDF reads the words in your scan using optical character recognition and saves a copy that looks identical but is searchable and copyable. Recognition runs in your browser, so the scan stays on your device; the language data for your chosen language is downloaded once, the first time you use it.

Pick the document's language before you start — English, Spanish, French, German, Portuguese, Italian and Dutch are available — because the language model is what turns shapes into the right letters.

The quick answer

1. 1Open the OCR PDF tool and upload the scanned document.
2. 2Choose the language the document is written in.
3. 3Run OCR, wait while each page is recognised, and download the searchable PDF.

How to tell whether you need OCR

Open the PDF and try to select a line of text with your cursor. If words highlight, there is already a text layer and OCR would add nothing. If your cursor draws a rectangle over the page instead, the page is an image and OCR is the step you are missing. The same test explains most complaints about empty PDF to Word conversions.

What OCR actually does

Recognition happens page by page. Each page is rendered as an image, the engine finds lines and word shapes, compares them against a trained model for your language, and produces text with positions. Those words are then written into the PDF as an invisible layer aligned with the picture of the page. That is why the result looks unchanged: you are still seeing the scan, with searchable text sitting behind it.

Input quality decides accuracy

• Scan or photograph at a decent resolution — small, blurry letters are the main cause of errors.
• Keep pages straight. Skewed and curved lines confuse line detection.
• Use even lighting and avoid shadows across the page when photographing.
• Straighten sideways pages with Rotate PDF before running OCR; upside-down text recognises poorly.
• Avoid heavily compressed scans, where letters have already lost their edges.

Choosing the right language

The language setting is not cosmetic. A model trained on English will misread accented characters, and a Spanish model will stumble over German compounds. Pick the language the body text is written in — that matters more than the odd foreign name or heading. For genuinely mixed documents, consider splitting the file by language section with Split PDF and running OCR on each part.

What OCR is good for

• Finding a contract clause or invoice number inside a folder of old scans.
• Copying a paragraph out of a scanned book or report instead of retyping it.
• Preparing a scan for PDF to Word, so the .docx contains real text.
• Making archived paperwork usable years later, when nobody remembers what is in which file.

Realistic expectations

OCR is very good, not perfect. Clean printed text usually recognises with few errors; handwriting, decorative fonts, stamps and faint carbon copies are much harder. Always proofread numbers before relying on them — a misread 8 in an invoice total causes more trouble than a missing full stop. Layout is not reconstructed either: the text layer follows reading order, so tables become sequences of words rather than grids.

Conclusion

Check for a text layer, straighten and choose the right language, then let OCR run. What you get back is the same document you scanned, only now it can be searched, copied and converted — which is what makes an archive genuinely useful.