Letters are boxes. Or wingdings symbols. Or random Unicode that means nothing in your language. The PDF clearly has text but it's gibberish.
What's actually going wrong
Font embedded incorrectly. The font data in the PDF is corrupt or incomplete. Wrong character map. The font is fine but the mapping from displayed glyph to character is broken. Missing fonts with no fallback. The viewer can't find the referenced font and substitutes wingdings.
The fix usually requires rebuilding the text layer, since the file's font data is broken at the bytes level.
The quick fix
Run convert PDF to Word. Flint's OCR reads the visual output — which is correct even if the underlying text is garbled — and generates new clean text. Open in Word, confirm it looks right, export back to PDF.
The rebuilt PDF has fresh fonts and clean text. Display works everywhere.
If that didn't work
If OCR shows the same garbage as the original (meaning the visual output itself is garbage — boxes instead of letters), the source PDF doesn't actually contain text. Get the source document and re-export.
For PDFs where you have the original source and just need to fix the corrupted export, re-export with 'Embed all fonts' enabled and using standard fonts (Helvetica, Times) rather than exotic typefaces.
Prevent it next time
Embed fonts at export. Use commercial fonts with proper embedding licences. Test PDF exports on different machines before distributing widely.
FAQ
Why does my PDF show wingdings instead of letters?
Font substitution failed catastrophically — your viewer couldn't find the referenced font and fell back to a symbol font. Rebuild with OCR or get source and re-export.
Can I install the missing font to fix it?
Sometimes. If the font is freely available and you install it, your viewer can render the PDF correctly. But the PDF only looks right on your machine — others without the font still see garbage. Embedding at export is the proper fix.
Will OCR recover garbled text?
OCR reads visual output. If the visual is garbled (boxes instead of letters), OCR can't recover what's not visible. If the visual is correct but the underlying text is garbled, OCR rebuilds clean.
Are font issues common?
Yes, especially with PDFs from old workflows or non-standard fonts. Modern PDF exports with 'embed all fonts' enabled rarely have font issues.
Font corruption needs a rebuild. OCR via Flint and produce a fresh, readable PDF.