PageFinderv4.2.11.22Sign in

Blog

How to find duplicate files across Box folders

Where duplicates come from in Box, how to find the obvious ones by hand, why renamed copies are the hard case, and what finding them by content looks like.

November 20, 2026 Β· 5 min read Β· PageFinder team

πŸ“ ContractsAgreement_final.pdfπŸ“ Clients / NorthwindNW-2024-signed.pdfπŸ“ Shared / Inboxscan0042.pdfπŸ“ Finance / 2024Invoice-20418.pdfsame contentsha256 Β· 3 copies

Every shared Box account accumulates copies. Not because anyone is careless, but because of how work happens: a file is downloaded to review, re-uploaded with comments, saved again under a clearer name, sent to a client folder and to the matter folder, and restored from an email attachment when nobody could find it. Two years later there are five copies of the same agreement, three of them with different names, and nobody is sure which is the signed one.

Why it matters more than the storage

Storage is cheap; confusion is not. Duplicates cost time when someone opens the wrong version, they cost risk when a retention rule deletes one copy and misses four, and they make any count of "what we have" wrong. A content audit of a drive that does not account for duplicates overstates everything.

Finding the obvious ones by hand

Box's search will find files that share a name, so the first pass is simple: search for the names that tend to be copied β€” "final", "signed", "v2", the client's name β€” and look at the results side by side. Sort a folder by name and the (1) and copy suffixes line up. For a folder of a few hundred files this takes an hour and catches the copies that look like copies.

As of this writing, Box does not offer a duplicate report in its standard interface, and the Box Admin Console's reports are about activity and access rather than content. There are third-party tools that scan a Box account for duplicates; most compare names and sizes, some compare file hashes, and they generally need an administrator to authorise them for the whole account.

The hard case: same file, different name

A hash comparison catches a file that was uploaded twice unchanged, whatever it was renamed to, because the bytes are identical. That is the case that name searching misses entirely: Agreement_final.pdf in the matter folder and NW-2024-signed.pdf in the client folder are the same file, and nothing about their names says so.

πŸ“ ContractsAgreement_final.pdfπŸ“ Clients / NorthwindNW-2024-signed.pdfπŸ“ Shared / Inboxscan0042.pdfπŸ“ Finance / 2024Invoice-20418.pdfsame contentsha256 Β· 3 copies
The same file under three names in three folders, found by its content hash

What a hash comparison still misses is the near copy: the same agreement saved once as a Word file and once as the PDF of it, or scanned twice on different days. Those are not duplicates to a computer and are duplicates to a person. There is no automatic answer for near copies yet that does not involve reading the documents.

Finding them by content, across the whole account

PageFinder takes the hash approach across everything in the folders you connect, and counts the result on a dashboard: how many distinct documents you really have, how many identical copies exist and where, whatever they were renamed to. It reads Box read-only and in place, so nothing is moved, and the count is there as soon as a folder is indexed. Because it also catalogues each file β€” kind, date, parties, summary β€” the near copies tend to surface too: two "agreements" about the same parties and the same date, one Word and one PDF, sit next to each other in the catalog. See it on one folder: PageFinder for Box.

What to do with the list

Do not delete anything in bulk. Decide which copy is canonical β€” usually the one in the folder the matter or the client is organised by β€” and replace the others with a Box link to it, or leave them and note the canonical one in the catalog. The goal is that the next person finds the right file, not that the storage bill goes down by a few cents.