Your shared B-roll drive is full again. You know there's junk on it (abandoned downloads, outtakes, the same drone shot in four codecs) but every time you try to clean it up you get cold feet. There are ten active projects pulling from this drive and you have no idea which clips they're actually using.

This is the fundamental headache of shared asset libraries. The thing that makes them efficient (one file, many projects) is the same thing that makes them terrifying to maintain.

Why Shared Drives Rot Faster Than Project Drives

A project drive is at least bounded. One job, one footage set, one timeline. When the job wraps, the whole thing can be archived as a unit.

A shared B-roll drive has no natural boundary. Footage gets added when it's shot. It gets used by whichever projects need it. It never gets removed because removing it might break something. So it grows. Forever.

The result, in every doc shop and agency edit room I've seen, is a 20 to 40 TB volume where probably half the contents haven't been touched in eighteen months and a meaningful slice is genuinely orphaned. Not referenced by any current project, not part of any deliverable, just dead weight nobody is willing to delete because nobody can prove it's safe.

The Failure Mode of Manual Cleanup

The instinct is to clean up by date. Sort by "last modified," anything older than two years goes. Don't do this. Modification dates on a shared drive mean nothing. A clip from a five-year-old shoot might be cut into a sequence you delivered last week. The timeline doesn't care when the file was shot.

The other instinct is to clean up by folder. "We don't work with that client anymore, delete their footage." Also dangerous. Cross-project reuse means a clip from a dead client's drone footage might be the establishing shot in a current pitch reel.

Both methods fail for the same reason. They look at the file's metadata instead of asking the only question that matters, which is whether any active project's timeline references this file.

What "Active" Actually Means on a Shared Drive

Think of the shared drive like a rental house cage. A camera body in the cage is "active" if any current job has it on the gear list, not because it shipped recently or because someone remembers using it last quarter. Same logic applies to your assets.

The only safe definition of an active asset is: a file referenced by at least one .prproj somewhere on the system, on any timeline, across any project. Not "in any bin." Not "imported once." On a timeline, in a sequence, in a project file that still exists.

To know that, you have to do three things at once:

  1. Enumerate every .prproj file on every drive your team uses (not just the shared B-roll drive, the project drives too, because that's where the .prproj files actually live).
  2. Walk each project's reference chain from sequences down through SubClips and MasterClips to the actual paths on disk.
  3. Cross-reference those paths against the contents of the shared drive.

Anything on the shared drive that doesn't appear in step 3 is genuinely safe to remove. Anything that does, even once, stays.

This is a tedious cross-reference exercise if you do it by hand. For a 30 TB drive feeding twenty active projects, you're looking at days of work and a real chance of human error. But the underlying logic is the only logic that's actually safe.

The Workflow That Doesn't Break Things

Here's the cleanup pass I'd run on any shared B-roll drive:

Step 1: Freeze the drive. Tell the team no new imports for an hour. You don't want a project pulling in a file mid-analysis and then you marking it as orphaned.

Step 2: Inventory the project files. Locate every .prproj that any editor on the team is currently working on. Don't trust your own memory. Ask everyone, and check the recent files lists. Missing one project means falsely deleting whatever it depends on.

Step 3: Build the usage map. For each .prproj, parse the reference chain and produce a list of every absolute media path it depends on. Combine those into one master list.

Step 4: Diff against the drive. Walk the shared drive and flag every file that doesn't appear in the master list. That's your candidate set.

Step 5: Sanity check before deleting. Spot-check the largest candidates. Make sure nothing on it is something a producer might want even if no editor cut with it. Stock plates, brand assets, signed releases tied to footage. Cleanup is a one-way operation.

Step 6: Move, don't delete. Move the candidate files to a quarantine folder or a slow-tier drive for thirty days before you actually rm them. If nothing breaks in that window, they're gone for good.

The reclaim on a properly maintained shared drive is usually 30 to 60 percent of total capacity on the first pass. On a 30 TB drive that's 9 to 18 TB you don't have to buy. Run the numbers in the calculator for your storage tier and the savings are not subtle.

The Tool Question

Doing this by hand once is doable. Doing it quarterly, which is what a healthy shared drive needs, is not realistic for most teams. The cross-reference is too tedious and the cost of one mistake is too high.

This is the workflow Clip Sweeper automates. Point it at the project drives and the shared B-roll volume, and it builds the usage map across everything in one pass. Same logic outlined above, just without the spreadsheet. The output is a list of files no project anywhere depends on, with the reclaimable size attached.

Before authorizing another expansion of the shared drive, it's worth running a cross-drive analysis once and seeing what's actually load-bearing. The reclaim usually pays for itself before the next storage purchase even comes up.