You hit Ctrl+F, type your query, watch '0 of 0' come back. But the word is right there on the screen.
The PDF has no text layer. The page is essentially a picture, and Ctrl+F has no text to find.
What's actually going wrong
PDFs that are searchable have a vector text layer underlying the visible content. Scanned PDFs and some image-export PDFs lack this layer — there's no real text, just pictures of letters.
Adding OCR re-derives the text from the visual content and inserts it as an invisible layer behind the existing image. Search then works because there's actual text to find.
The quick fix
Run the file through convert PDF to Word. Flint's OCR reads every page and generates real text. Verify in Word, then export back to PDF.
The new file looks identical but has a full searchable text layer. Ctrl+F now works everywhere.
If that didn't work
OCR accuracy depends on source legibility. Sharp 300dpi scans recognise near-perfectly. Faded or low-res scans produce errors. For mission-critical documents, proofread the OCR output.
For PDFs that have a text layer but search still doesn't find specific terms, the layer might use unusual encoding or character maps. Re-OCR via Flint to rebuild cleanly.
Prevent it next time
Enable OCR-on-capture in your scanner settings — every scan arrives searchable. For exported PDFs, use 'standard' or 'high quality' export presets which preserve vector text.
FAQ
How can I tell if a PDF is searchable?
Try selecting text. If it highlights, you have a text layer. If selection doesn't work, the PDF is image-only and search won't find anything.
Does adding OCR change the PDF visually?
No. OCR adds an invisible layer behind the existing image. The visible content is unchanged; you just gain searchability.
Why is search finding only some words but not others?
OCR errors on specific words mean those weren't recognised cleanly. Re-OCR with higher source quality if possible, or accept some imperfection in low-quality scans.
Can I search Asian or non-Latin scripts?
Yes, Flint's OCR supports many scripts. Recognition quality varies by script and source quality but for standard documents in major languages, search works well.
Searchable PDFs need a text layer. OCR via Flint and Ctrl+F starts finding things again.