Compare troubleshooting: password errors, scans with no text diff, and all-red overlays

Compare runs entirely in your browser — neither PDF is ever uploaded — so most problems come down to the files themselves or which view fits them. Here are the common ones and how to fix them.

"…is password-protected. Remove the password first, then re-drop it here."

Cause: The PDF requires a password just to open, so Compare can't read its pages. (PDFs that only restrict editing or printing usually open and compare normally.)

1. Click the Remove the password first → link in the error, or open Unlock directly.
2. Enter the PDF's password to produce an unlocked copy — this also happens locally; the password and file stay on your device.
3. Re-drop the unlocked copy into its slot. If both files are protected, unlock each one.

"Couldn't load … — it may be corrupted or not a valid PDF"

Cause: The file isn't a well-formed PDF — a truncated download, an email-mangled attachment, or another format renamed to .pdf.

1. Confirm the file opens in another PDF viewer. If it doesn't open anywhere, re-download or re-export it from the source.
2. If it's actually an image or a Word file, convert it to a real PDF first, then compare.
3. If it opens fine elsewhere but still fails here, report it — that's a case the team wants to see.

Text diff says "No text differences on this page" but the pages clearly differ

Cause: One or both pages have no text layer to read. Scanned PDFs are photographs of paper — to the text diff, two visibly different scans can both look like empty pages. The same applies to pages where the "text" is part of an image, such as flattened or rasterized exports.

1. Switch to the Overlay diff view. It compares rendered pixels, not text, so it works on scans — every region that differs gets a red wash, with a "% of pixels differ" figure.
2. If you need a word-level redline, run each scan through Searchable (OCR) first to add a selectable text layer, then compare the OCR'd copies.

Overlay diff paints almost the whole page red

Cause: The pixel comparison flags every position where the pages render differently — and a small change can move everything after it. Common triggers: text that reflowed after an early edit, pages scanned at a slight skew or offset, different page sizes, or one version exported with different margins.

1. Switch to the Text diff view — it matches words by content, not position, so a paragraph that merely moved down the page isn't flagged.
2. For scans, check the "% of pixels differ" figure in context: a skewed rescan of an identical page can legitimately show a high percentage. The side-by-side view is often the saner check for scan pairs.
3. Remember which side is which: the overlay draws page A and washes red where B disagrees. Use Swap if you loaded them in the wrong order.

Text diff marks the entire page as removed and re-added

Cause: A deliberate safety valve. On very long pages the word-level matcher would freeze the tab, so Compare falls back to showing the whole old page as removed and the whole new page as added instead of hanging. The cutoff depends on both pages together — when the two versions are similar in length it works out to roughly two thousand words per page, and one extremely long page can trip it even if the other is short.

1. Use Side-by-side or Overlay diff for those pages.
2. Or split the documents into smaller pieces with Pages and compare the relevant sections — pages up to roughly two thousand words get the full word-level diff.

Page navigation stops before the end of one document

Cause: The two PDFs have different page counts. Compare pairs pages one-to-one, so it only walks the first N pages, where N is the shorter document's count — the header says so (for example "A has 4, B has 6 — comparing first 4"), and a note under the viewer names which file has extra pages past the comparison range.

1. To check the extra pages, open that file on its own in Pages or any viewer.
2. If the versions drifted because pages were inserted mid-document, comparing page-by-page will misalign everything after the insertion — extract matching page ranges with Pages first, then compare those.

"Couldn't render a page for comparison" or "Couldn't read text for the diff"

Cause: One of the PDFs is damaged mid-file or uses a feature the renderer can't decode — earlier pages may have worked fine.

1. Re-export or re-print the file to PDF from its source application; that usually rebuilds the broken structure.
2. Try the other views — a page that fails text extraction may still render for the overlay, and vice versa.

The comparison disappeared after a refresh

Cause: Not a bug — both documents live only in your browser's memory, never on a server, so a refresh genuinely clears them. The browser shows a leave-page warning while files are loaded to prevent exactly this.

1. Re-drop both files; loading is quick since nothing re-uploads.
2. If you need the differences on record, note them down or screenshot the views — Compare is a viewer and doesn't export a marked-up file.

Still stuck?

If none of these match what you're seeing, contact support and describe what happened — since your files never leave your browser, support can't see them, so include each file's size and page count, which view you were in, and the exact error text.