How to tell if a PDF is searchable, and make it so
Press Ctrl+F and search for a word you can see on the page. If it is not found, try to select the text with your mouse. If the cursor only draws a rectangle instead of highlighting words, the PDF is image-only and needs OCR to become searchable.
A PDF that will not search is one of the most common document complaints, and it has three quite different causes. Two minutes of testing tells you which one you have, and the fix follows directly from that.
Test 1: the Ctrl+F test
Open the file in any viewer, press Ctrl+F (Cmd+F on a Mac), and type a common word you can plainly see on the current page. Something short and unambiguous: a name, a date, "the".
No results means there is no matching text layer. But do not stop here, the search box may simply be searching a page you are not looking at, and some viewers restrict search scope in odd ways. Run the second test to confirm.
Test 2: the select-text test
Click at the start of a line and drag across it. This is the decisive test.
- Words highlight as you drag, following the shape of the line: there is a real text layer. Your search problem is something else.
- A blue rectangle is drawn over the page and no words highlight: the page is an image. This is an image-only PDF and needs OCR.
- Text highlights but pastes as gibberish: there is a text layer, but the font encoding is broken. Different problem, covered below.
Test 3: the file size test
A useful sanity check before you even open the file. Divide the file size by the page count.
| Size per page | Almost certainly | Searchable? |
|---|---|---|
| 10 - 80 KB | Text and vector graphics, born digital | Yes |
| 80 - 300 KB | Text plus a few photos | Yes |
| 300 KB - 1 MB | Colour scan, JPEG compressed | Only if OCR was applied |
| 1 - 5 MB | High-resolution or lossless scan | Usually not |
| Under 10 KB | Bilevel fax-style scan, or a nearly blank page | Usually not |
On the command line, Poppler settles it in one line: pdftotext file.pdf - prints the extracted text to your terminal. Empty output, or output consisting only of form feeds, means there is nothing to extract.
Fixing an image-only PDF
This is the common case and the fix is straightforward: run OCR, which adds an invisible text layer over the existing images without changing how the page looks.
Add a searchable text layer to a scanned PDF in your browser, in over 100 languages.
OCR PDF →The step-by-step version, along with language selection and the accuracy you should expect, is in how to OCR a scanned PDF. If the scan is poor, spend five minutes on improving the input before running recognition, it makes far more difference than any setting.
The awkward middle case: hybrid PDFs
Documents assembled from several sources are frequently part searchable and part not. A born-digital report with three scanned appendices behaves normally for forty pages and then goes silent. So does a contract where a signed page was scanned and reinserted.
The symptom is distinctive: search works, but misses things you know are there. Test the specific page you care about rather than page one. Most OCR tools have a skip-text option that processes only the image pages and leaves the existing text layers untouched, which is exactly what you want here, re-OCRing a page that already has good text usually makes it worse.
Text that selects but pastes as nonsense
You highlight "Invoice total" and paste it, and get something like ,QYRLFH WRWDO. The page has a text layer, but the PDF is missing the mapping that says which Unicode character each glyph represents.
- A subset font with no ToUnicode map. The generator embedded only the glyphs it used and did not include the translation table. Very common with older LaTeX output and some report generators.
- Ligatures. "fi" and "fl" are single glyphs in many fonts, so words like "define" lose characters on copy even when everything else is fine.
- A symbol or custom font. Icon fonts and some corporate typefaces map letters to arbitrary code points by design.
There is no clean repair for this from outside the file. The practical workaround is to convert the pages to images and OCR them, which recognises the visible shapes and ignores the broken encoding entirely. It is a heavy fix, but it is reliable.
When search still fails on a text PDF
| Symptom | Likely cause | Fix |
|---|---|---|
| Search finds nothing anywhere | No text layer at all | Run OCR |
| Search works on some pages only | Hybrid document | OCR with a skip-text option |
| Text selects but copies as gibberish | Missing ToUnicode map | Rasterise those pages and OCR them |
| Search misses accented or non-Latin words | Wrong OCR language, or encoding loss | Re-run OCR with the correct language |
| Search misses hyphenated words | Line-break hyphens are in the text layer | Search for the shorter fragment instead |
| Viewer search is slow or incomplete | Very large file or no index | Extract with PDF to text and search that |
A searchable PDF is not an editable one. OCR gives you findable words sitting invisibly over a picture; the picture is still what you see. To actually edit the content you need to convert it to Word, which rebuilds the layout as real paragraphs and accepts that some formatting will shift.
Frequently asked questions
How can I tell if a PDF is searchable?
Try to select a line of text with your mouse. If individual words highlight, there is a real text layer. If the drag only draws a rectangle over the page, the document is image-only and needs OCR.
Why does Ctrl+F find nothing in my PDF?
The most common reason is that the pages are scanned images with no text layer. It can also happen on hybrid documents where only some pages were scanned, or when the fonts lack a Unicode mapping.
Why does copied text from a PDF come out as gibberish?
The embedded font has no ToUnicode map, so the viewer knows which glyph to draw but not which character it represents. Rasterising those pages and running OCR over them recovers the text reliably.
Can a PDF be partly searchable?
Yes, and it is common. Documents assembled from a digital report plus scanned appendices search normally on some pages and not at all on others. Use an OCR tool with a skip-text option so existing text layers are left alone.
Does making a PDF searchable make it editable?
No. OCR adds an invisible text layer over the scanned image; what you see is still the picture. To edit the content, convert the document to Word or another editable format and expect some layout to shift.