Duplicates and reclaiming space
You'll need: an administrative role to open the screen, and media delete permission to act on a group, since resolving duplicates deletes files. Administration → Media Duplicates.
Libraries accumulate copies: the same photo uploaded twice, a resized version saved alongside the original, a file re-exported with a new name. They consume storage and they make searches ambiguous.
Find duplicates
The duplicates screen groups files that are identical or near-identical and shows each group with a suggested keeper, usually the largest, earliest, most-used copy.
The suggestion is a starting point, not a decision. Check it against the group's actual contents before acting: the highest-resolution file is not always the one your designs point at.
Before you delete anything
The screen shows what depends on each file. A copy used in three designs and a share link cannot be removed without breaking them.
Removing a duplicate re-links whatever was using it to the keeper first, then deletes. That ordering is the reason to do this here rather than by hand in the library, where deleting leaves designs and posts pointing at nothing.
Reclaim space
Administration → Storage is where you see what the library is actually using:
- what is stored, broken down by module, file type, origin and person
- the largest files
- how much is reclaimable, from duplicates and from the trash
Two buttons on that screen do the reclaiming: Review duplicates, and Open trash.
The trash
Deleted files stay recoverable for a retention window set by your storage plan, under Media → Library → Recently Deleted. They still count against your storage until that window passes.
Each shows when it was deleted and how long it is recoverable for. Recover puts it back with its tags, albums and analysis intact.
If a file shows Not recoverable, the underlying object is no longer retained and recovery is not possible.
Emptying the trash is the fastest way to free space and the only irreversible one on this page. Everything else here can be undone.
Prevent duplicates
One habit prevents most of this: when you have a corrected or re-exported version of a file already in the library, add it as a new version of that file rather than uploading it as a new one. See uploading files.
Versions count against storage too, and the library keeps a bounded number of them per file. But they stay attached to one identity, so the library never has to guess which of four files is current.