It’s a small, maddening moment: you open a scanned PDF, you can plainly read the words on the page, you press ⌘F, type one of those exact words — and Preview reports nothing. It feels like a bug. It isn’t. Understanding why takes about two minutes, and it explains a whole family of related mysteries: why Spotlight ignores your scans, why you can’t select text in some PDFs, and why “PDF” can mean two completely different things.
A PDF is a container, not a format for text
PDF was designed to reproduce a page’s appearance exactly, anywhere. To do that, the format is a container that can hold different kinds of content:
- Text objects — actual character codes plus fonts and positions. This is what a PDF exported from Word, Pages, or a web browser contains. Software can read the characters directly, which is why search, selection, and copying work.
- Image objects — pictures, placed on the page. A photo in a brochure, a logo — or an entire page, if the page came from a scanner.
When you scan paper, the scanner captures a photograph of the sheet and wraps it in PDF packaging. The result opens in any PDF viewer and prints beautifully, but inside there are no characters at all — just a grid of pixels that happens to look like characters to a human. When you press ⌘F, Preview searches the file’s text objects, finds none, and truthfully answers: nothing here.
You read the page with your visual system, which effortlessly turns shapes into words. The computer, without OCR, never performs that step. You and the software are looking at the same page through different machinery — which is why you’re both “right.”
How to tell which kind of PDF you have
- The selection test. Try to select a few words. Real text selects word by word, with the highlight hugging each line. An image-only page either selects nothing or draws a rectangular marquee, as if you were cropping a photo.
- The search test. ⌘F for a word you can see. Found → text. Not found → image (or terrible OCR — see below).
- The zoom test. Zoom way in. Real text stays razor-sharp at any magnification because it’s drawn from fonts. Scanned text goes soft and pixelated, because it’s a picture.
One wrinkle: Live Text on modern macOS can let you select and copy text from a scanned page in Preview, on screen. That’s Apple’s on-device recognition running in the moment — genuinely useful, but it writes nothing back into the file. The document on disk still has no text objects, so ⌘F in other apps, Spotlight, and anything that reads the file still see a blank scan. Live Text makes the screen readable, not the file searchable.
The fix: give the file a text layer
OCR — optical character recognition — is the step your visual system does for free: reading shapes and deciding what characters they are. Good OCR tools then do something Live Text doesn’t: they write the recognized characters into the PDF as an invisible text layer, positioned precisely under the printed words on the page image.
The result is a hybrid: what you see is still the original scan, pixel for pixel; what software sees is real text in the right places. Search works, and highlights land on the printed words. Selection works. Copy-paste works. Spotlight indexes the file like any other document (how that works).
You have several ways to add that layer — Acrobat Pro, ABBYY FineReader, cloud services (with a significant confidentiality caveat), command-line tools, all compared in the full how-to guide. The way we built, because we wanted the fix to be as fast as the annoyance: right-click the PDF in Finder, choose Make Searchable (OCR), and RightClickOCR writes a new “Name (Searchable).pdf” beside the original — recognized entirely on your Mac by Apple’s Vision framework, with the original left untouched.
Related mysteries, same explanation
- “Spotlight can’t find my scans.” Spotlight indexes text content; image-only PDFs have none to index.
- “Some pages search fine, others don’t.” Mixed PDFs are common — a digital cover page stapled to scanned exhibits. The digital pages have text objects; the scanned ones don’t.
- “Search misses words I can clearly see.” The file may have a bad text layer from low-quality OCR of a poor scan — crooked, blurry, or fax-grade pages produce garbled recognition. Re-OCR-ing with a better tool, after dealing with skew, usually helps.
- “The PDF says it’s 2 MB for one page.” Page images are big; text is tiny. Large file size per page is a strong hint you’re holding a scan.
Once you know the two-kinds-of-PDF secret, the “Not found” message stops being maddening and becomes a diagnosis: this file needs a text layer. That’s a solved problem — solve it once per file and ⌘F behaves forever after. And because the text layer travels inside the PDF, the fix is permanent and portable: the searchable copy stays searchable when you email it to a colleague, open it on another machine, or file it away for five years. The scan you fix today is the search result you’ll be grateful for later.