How to find duplicate files without installing anything

20 August 2026 · 8 minute read

Between 30 and 50% of the space your documents and photos occupy is a copy of something you already have. That number is not a marketing estimate: it is the figure Microsoft publishes in its own documentation. And the curious part is that neither Windows nor macOS gives you a tool to find them.

How much space this really is

Microsoft publishes the typical savings of its deduplication technology by content type. For user documents — the Office files, photos, music and video of an ordinary person — the figure is 30 to 50%. For mixed work folders, 50 to 60%.

Translated: out of every 100 GB of your personal files, somewhere between 30 and 50 GB are probably repeats. On a 512 GB drive that is the difference between having room and deleting things every week.

Here is the awkward detail: that Microsoft feature only exists on Windows Server. It is not on the Windows running your laptop. It is not hidden in a menu — it simply does not ship.

Where all those copies come from

Nobody duplicates files on purpose. They pile up on their own, and almost always through the same five routes:

• Repeated downloads. You grab the same attachment three times and the browser leaves you invoice.pdf, invoice (1).pdf and invoice (2).pdf.
• Messaging. The same photo arrives on WhatsApp, you save it, somebody forwards it to the group, and you save it again. Now you have two versions at different quality.
• Half-finished backups. You copied "Desktop" to an external drive, then copied "Documents" which also held part of the desktop, and years later nobody knows what is inside Old_Backup_2019.
• Cloud sync. A sync conflict creates report (conflicted copy).docx, and there it stays forever.
• The "just in case". Duplicating a folder before reorganising it is sensible. Forgetting to delete the copy afterwards is human.

Why searching by name does not work

It is everyone's first instinct: sort by name and look for the ones that resemble each other. And it is exactly what works least.

The name is the part of a file that changes most easily and means least. The same photo can be IMG_2043.jpg on the phone, beach-holiday.jpg after you renamed it, and WhatsApp Image 2026-03-14.jpg when it came back from the family group. Three names, one set of contents.

And the reverse: two files with the same name can be completely unrelated. budget.xlsx in two different client folders are two different documents that it would be a disaster to confuse.

macOS shows the problem clearly. You can build a Smart Folder in Finder that looks for files with the same name, but it will not detect the ones named differently that are identical — which is the majority case. Finder ships no real duplicate finder.

How they are actually found

Serious detection works on three levels, each more expensive than the last. A well-built finder applies them in order so it does not waste time.

Level 1 · Size, as a fast filter

Two identical files weigh exactly the same. So if a file's size is unique across the whole folder — down to the last byte — it cannot have a twin, and there is no need to even open it.

It sounds like a minor detail and it is what separates a scan of seconds from one of half an hour. In a folder of video, reading every 4 GB file to compare it against all the others is unworkable; discarding by size removes most of them before they are touched.

Level 2 · The fingerprint of the contents

Files that do share a size get a SHA-256 fingerprint: a 64-character number summarising every byte in the file. If two files produce the same fingerprint, they are the same file. Change a single bit — a pixel, a comma, a frame — and the fingerprint changes completely.

This is what makes the name stop mattering. The fingerprint is indifferent to what a file is called, which folder it sits in, or when it was made. And it works just as well on a video, a song, a PDF or a spreadsheet, because it only looks at bytes.

Level 3 · Visual comparison, images only

That leaves the most common case with photos: same image, different files. The photo that came back from WhatsApp smaller and more compressed has different bytes, a different fingerprint, and is nonetheless the same photo.

There you have to compare appearance: each image is reduced to a tiny grayscale grid and the two are measured against each other. Done well, this recognises the same photo through a resize or a quality change. Done badly, it groups things with nothing in common — the classic failure with folders of screenshots, which all look alike once shrunk.

What your operating system ships with

Which is why an entire market of desktop programs exists for something that should come included.

The three ways to do it today

A desktop program

The traditional and most powerful option. The catch is what it involves: downloading an executable from a site you do not know, installing it with administrator rights, and trusting it carries no extras. Many of the free ones are funded by adware, and several cap deletion until you pay.

An online service that uploads your files

Convenient until you think about what you are uploading. To find duplicates you have to send the entire folder to somebody else's server: your family photos, your scanned documents, your contracts. No amount of reclaimed space is worth that.

The browser

This is the one almost nobody knows is already possible. Modern browsers allow a page, with your explicit permission, to open a real folder on your disk, read it and write back to it. That permission dies the moment you close the tab.

Which means a web page can do the whole job with nothing installed and without a single file leaving your machine. Your computer does the work; the site only supplies the code.

What you should never delete

This is the part almost no article covers, and the one that can genuinely cost you. Not every repeated file is surplus. There are folders where duplication is deliberate and necessary.

• System and program folders. On Windows, C:\Windows and Program Files are full of deliberately repeated libraries: each program carries its own copy of the version it needs. Deleting "the duplicate" breaks applications, sometimes days later.
• Development folders. node_modules, Python virtual environments, build caches. They are packed with identical files the project needs in that exact place.
• Managed photo libraries. Apple Photos, Lightroom catalogues, the iCloud library. These apps keep their own internal index; deleting files from outside leaves the catalogue pointing at nothing.
• Cloud-synced folders. If you delete a file inside your OneDrive, Drive or Dropbox folder, you are deleting it in the cloud and on every device too. Pause syncing before reorganising, or work on a copy.
• Crops and edited versions. A crop is not a duplicate. Delete the original and you lose pixels; delete the crop and you lose the framing you chose. Whichever you remove, you lose something.

And one recommendation that applies to every case: move instead of delete. Send the extras to a separate folder, live with it for a week, and delete it once you have confirmed nothing broke. It is an extra step that costs nothing and avoids the only irreversible mistake in this whole process.

How to stop them piling up again

• Empty Downloads every month. It is where half the duplicates on any machine are born.
• One home for photos. Pick a folder or an app and make it the only one. Most photo duplicates come from having three destinations and using them at random.
• Resolve sync conflicts when they appear. That "(conflicted copy)" file is never going to fix itself.
• Clean before a big backup. Duplicating a folder that already had duplicates multiplies the problem.
• Re-run the scan twice a year. It takes minutes and stops you ending up back where you started.

In short

Duplicates take up a fraction of the disk almost nobody imagines, your operating system ships nothing to find them, and the file name — the first thing all of us look at — is precisely the least reliable signal for spotting them.

What does work is comparing contents: size to discard fast, a fingerprint to confirm, and visual comparison for photos. And your own browser can do all of it today, with nothing installed and without your files going anywhere.

Sources

• Microsoft Learn — Data Deduplication Overview: typical savings by content type and Windows Server-only availability.
• Macworld — how to find and delete duplicate files on a Mac: the Photos "Duplicates" album and the limits of Finder Smart Folders.