How to make a scanned PDF searchable without changing how it looks
A 'searchable PDF' is the same scan you started with, plus an invisible text layer that makes the content findable through Cmd+F, Google indexing, document management systems, and any search workflow you build on top. Done right, the file looks pixel-identical to the scan but behaves like a text PDF. Done wrong, the text layer drifts or fails on accented characters and your search returns nothing.
What 'making it searchable' actually does
We OCR every page, then position each recognized word on top of the matching pixels in a hidden text layer. The layer is part of the PDF but renders invisibly — when you search for a word, the PDF reader matches against the hidden layer; when you copy text, the copy comes from the hidden layer; when a search engine indexes the file, it indexes the layer. Visually, nothing changes.
This is different from straight OCR-to-text. The output is still a PDF, the original page images are preserved, and the file's hash is essentially the same one that was archived — just with the text added as metadata.
When this is the right operation
Pick this tool when you specifically need the source PDF to remain visually intact:
- Archives that must preserve the scanned image. Court records, certified copies, regulatory submissions where the scan IS the document.
- Document management system uploads. DMS platforms index the text layer for full-text search across thousands of files.
- Internal compliance archives. Auditors search by keyword across decades of contracts.
- Searchable shared drives. Google Drive, SharePoint, and Box index searchable PDFs but not image-only ones.
- Long-term searchable storage of receipts and invoices. Find any expense by vendor name without flipping through scans.
Common mistakes
Patterns that make 'searchable' PDFs that aren't actually searchable:
- OCR'ing the wrong language. Auto-detect handles most cases, but explicitly setting the language eliminates the risk.
- Skipping deskew on tilted scans. Tilt confuses the layer alignment; the search highlight may land 50 pixels off the actual word.
- Compressing after making searchable. Strong compression downsamples the page image but the text layer stays put — small misalignments creep in. Make searchable first, compress on Light only.
- Forgetting that handwriting won't OCR. Most OCR engines (including ours) won't reliably OCR handwriting. The result for a handwritten page will be sparse or empty text. Use the AI OCR's handwriting model only when needed.
- Not testing search after the fact. Open the resulting PDF and Cmd+F a known word from the document. If it doesn't find, the OCR pass missed something.
Tips for clean results
Practical adjustments that improve searchability:
- Use the highest-DPI scan available. 300 DPI is the floor for reliable accuracy. 600 DPI is overkill but doesn't hurt.
- Set the language explicitly when bilingual. If the document mixes English and Spanish, list both. Auto-detect picks one and may miss tokens in the other.
- Preserve original PDF metadata. Author, created-date, and document-id should pass through. Toggle 'preserve metadata' on by default.
- Run on a copy. Always preserve the unsearchable original alongside the searchable output. Re-OCR is cheap; re-scan isn't.
- Bundle with PDF/A if archival. PDF/A-2u is the searchable archival standard — the ISO requires the text layer to be present and Unicode-mapped. We produce PDF/A-2u when you toggle it on.
Limits and compatibility
Free tier: up to 20 pages free, then 2 credits per page. Output is a standard searchable PDF that opens in every reader (Acrobat, Preview, every browser, every DMS). The hidden text layer is recognized by Spotlight, Windows Search, Google Drive, and OCR-aware platforms. Supports 30+ scripts and most Latin-script languages out of the box.