Why Can't I Select Text in My PDF? Scans, Text Layers and OCR

July 24, 2026 · PDF Tools

Open two PDFs side by side. In one, you can select a sentence, search for a word, copy a paragraph. In the other the cursor selects nothing, search finds nothing, and copying gives you nothing — even though the page is clearly full of words. Both are valid PDFs. The difference is what is inside them, and it determines what any tool can do with the file.

Digital PDFs carry text; scanned PDFs carry pictures

A PDF exported from Word, a browser or an invoicing system contains real text: character codes, the font to draw them with, and coordinates for where each run of characters sits on the page. Your viewer draws the letters on demand, which is why they stay razor-sharp at any zoom, why search works, and why the file is small.

A PDF from a scanner or a phone camera contains none of that. Each page is one large image — millions of pixels that happen to depict words. To the file format, a scanned contract and a scanned photograph of a cat are the same thing: a picture. There is nothing to select because, structurally, there are no letters present at all.

The quick test

Try to select a line of text in your PDF viewer, or press Ctrl+F (Cmd+F on a Mac) and search for a word you can see on the page. Selection works and search finds it: digital text layer. Nothing selects and search comes up empty: a scan. Zooming is a second clue — digital text stays perfectly crisp at 800% while scanned text dissolves into pixels.

Why this matters for every PDF tool

• Text extraction reads the text layer. On a digital PDF it returns exactly what the document contains; on a scan it correctly returns nothing, because nothing is there.
• Compression behaves oppositely: scans compress dramatically (they are all image data), while clean digital PDFs barely shrink (text is already tiny).
• Search, accessibility and copy-paste all depend on the text layer. A scan is invisible to document search and to screen readers.
• Structural operations — merge, split, rotate, reorder, watermark — work identically on both, because they treat pages as units without caring what is on them.

OCR: turning pictures of words back into words

Optical character recognition looks at the image, recognises letter shapes, and writes an invisible text layer positioned behind the picture — making the scan searchable and selectable while it still looks identical. Modern OCR is excellent on clean printed pages and English-like scripts, less reliable on handwriting, unusual fonts, low-resolution scans and complex tables. It is genuine recognition, so unlike extraction it can be wrong: a smudged 8 can come back as a 3. OCR is computationally heavy and needs trained recognition models, which is why it is a server-side feature everywhere you find it, and why our browser-based extractor does not attempt it.

Practical advice

1. Need the text out of a scan? It needs OCR first — extraction alone cannot conjure a text layer.
2. Producing documents yourself? Export to PDF directly from the application instead of printing and scanning; the result is smaller, searchable and accessible.
3. Stuck with a scanner? Scan at 300 DPI if you plan to OCR the result later — recognition accuracy drops sharply below that.
4. Verify important numbers after OCR. Recognition is probabilistic, and a bank statement deserves a glance at the totals.

Try it yourself

• Extract Text & TablesPull text as .txt and detected tables as .csv from a PDF.
• Compress PDFShrink a PDF's file size with selectable quality levels.
• Organize PagesRotate, reorder and delete pages with drag-and-drop thumbnails.