Searchable (OCR) troubleshooting: encrypted PDFs, no text found, slow runs, and the page cap

Common problems when making a scanned PDF searchable at Make searchable, and how to fix each one. Remember that OCR runs entirely in your browser — your PDF is never uploaded — so your device's CPU and memory do all the work.

"This PDF is encrypted" and the Make searchable button is disabled

Cause: the PDF requires a password to open. AttachKit probes the file as soon as you drop it, before any OCR starts, because the engine can't read password-protected pages.

Fix:

1. If you know the password, open Unlock, remove the password there, and download the decrypted copy.
2. Alternatively, open the PDF in the app it came from (Preview, Acrobat, etc.) and re-export it without a password.
3. Drop the unprotected copy back into Make searchable.

"This file isn't a valid PDF"

Cause: the file is corrupted, truncated (often from an interrupted download), or isn't actually a PDF despite the extension.

Fix:

1. Re-download or re-export the file from its original source.
2. If it came from a scanner, re-save or re-scan to a fresh PDF.
3. Drop the new copy into the tool.

"This PDF has N pages — too many to OCR in one pass"

Cause: Make searchable caps a single run at 200 pages. OCR costs 5–60 seconds per page, so a several-hundred-page scan would pin your browser tab for a very long time and risk running out of memory.

Fix:

1. Open Pages and split the document into parts of 200 pages or fewer.
2. Run each part through Make searchable separately.
3. If you want one file back, merge the searchable parts again with the Pages tool.

"OCR didn't find any text"

Cause: either the pages have no readable content (blank, extremely faint, or heavily distorted scans), or the selected OCR language doesn't match the document.

Fix:

1. Check the OCR language dropdown — it must match the language the document is written in. The tool remembers your last choice, which may be left over from a previous document.
2. Zoom into the PDF in any viewer and confirm a human can actually read the text. If the scan is illegible to you, it's illegible to OCR; re-scan at a higher resolution (around 300 DPI) if you can.
3. Run Make searchable again.

Recognized text is garbled or full of mistakes

Cause: the wrong OCR language, a low-quality scan, or unusual fonts/handwriting. OCR is statistical — even good runs make some mistakes, and handwriting is generally beyond it.

Fix:

1. Set the correct language (24 are available) and re-run — this is the most common cause.
2. If the source is a photo or fax-quality scan, re-scan flat at around 300 DPI with even lighting.
3. Accept that a few errors are normal: the visible page is untouched (the text layer is invisible), so a misread word only affects search and copy-paste, never how the document looks. Very rare characters outside the embedded font's coverage are stored as ?.

OCR is very slow, or you can't wait for it to finish

Cause: all recognition happens on your device — 5–60 seconds per page is normal, and large, dense, or high-DPI scans sit at the slow end.

Fix:

1. Watch the progress panel: it shows the stage, the current page ("Page X of Y"), and a time estimate once a few pages are done.
2. Keep the tab open and in the foreground while the run is going.
3. If you need to stop, click Cancel OCR. You'll see "OCR cancelled. Progress saved" — every completed page was checkpointed to encrypted storage on your device.
4. Return to the tool later and use Resume from where you left off on the home screen; the run continues from the next unfinished page instead of starting over. The same applies after a tab crash or accidental close — drafts are kept for up to 30 days (and cleared as soon as the finished PDF downloads), and you can delete them from the same list.

The language won't load, or OCR fails right at the start

Cause: the first use of a language downloads about 5 MB of training data into your browser cache. English and Russian are served directly from AttachKit; every other language fetches its training data (plain data, not executable code) from a public CDN, which can fail offline or on a network that blocks CDNs.

Fix:

1. Check that you're online and retry — once downloaded, the data is cached for future runs.
2. If your network blocks third-party CDNs, try English/Russian (self-hosted) or run the tool from a different network.

Still stuck?

If none of this fixes it, contact support and describe what you tried — but keep in mind nobody at AttachKit can see your document, because it never left your browser.