Start with an illustrative five-file batch
Imagine a folder with five files: contract-final.pdf, a byte-for-byte backup named contract-copy.pdf, a separately saved contract-signed.pdf with the same extracted words, a scanned contract-scan.pdf, and an unrelated invoice.pdf. Select all five in PDF Duplicate Review. The extension reads the chosen files locally and never deletes, renames, or rewrites them.
Keep byte identity separate from text candidates
The first result answers the narrow question: which files have identical bytes? A SHA-256 match makes contract-final.pdf and contract-copy.pdf exact copies even though their names differ. The second result compares normalized extracted text page by page. It may group contract-signed.pdf as a review candidate, but signatures, annotations, images, form values, and layout can still differ. A text match is evidence to inspect, not permission to delete.
Treat scans as a separate review queue
A scanned PDF may contain only page images. PDF Duplicate Review does not perform OCR or image comparison, so contract-scan.pdf may be reported as having no extractable text. Keep it for manual inspection or process it with a suitable OCR workflow before comparing text. Do not assume that an empty text result means the document is blank or unique.
Export the CSV before changing files
Download the CSV report and keep it with the review notes. It records candidate groups, the comparison basis, skipped files, and errors. Open every candidate side by side, confirm the document purpose and retention requirement, and record which copy should remain. Delete or archive a file only after that manual review. The extension deliberately provides evidence rather than an automatic cleanup action.
Use the batch size that fits the job
The free plan handles up to 10 PDFs per batch, including both comparison signals and the CSV report. Pro for version 1 costs EUR 49 once, tax included, raises the batch limit to 250 PDFs, and supports up to three active browser installations. Each Pro batch needs an online license check; PDF contents and comparison data are not sent with that check.
Common questions
Does the extension upload PDFs for comparison?
No. Selected PDF contents, filenames, extracted text, and hashes are processed locally in the extension.
Can a matching-text result prove two PDFs are identical?
No. Only the byte-identical signal proves that the selected files have identical bytes; text candidates still need manual review.
Practical documentation for focused software products with explicit privacy, licensing, and operating boundaries.