Cleaning a personal drive is simple. You know what you put on it, you know what you stopped using, and the only person you can break a project for is yourself. Cleaning a shared NAS where six other editors might depend on a file is an entirely different problem.

Think of the post-house NAS like the gear cage at a rental house. Anyone can pull from it. Most things are labeled. Some things are not. Toss the wrong cable and the next shoot finds out at exactly the wrong moment. Same logic applies to a shared edit server.

Here is the procedure that has cleaned hundreds of terabytes of dead weight off post-production NAS arrays without anyone getting paged at midnight.

The "Who Used What" Problem

Before you delete anything, you need to know who depends on what. The honest answer is: you cannot fully know. But there are tools, and some are better than others.

Server access logs tell you when files were read, but they do not tell you whether the file is referenced by an active project. A read three months ago might be the last time anyone touched the file, or it might be the moment the editor finished the cut and never reopened the project. The log does not know the difference.

File modification time (mtime) lies. A camera-original RAW might have an mtime of three years ago because nobody re-encoded it. That file is still the source for a final that ships next week. Mtime tells you when the file was last written, not when it was last needed.

"Archive everything" is the wrong default. It feels safe but it just moves the problem. Archive drives fill up too. Within two years, the archive drive is full of more abandoned projects than the live NAS was when you started.

The only correct way to know what is in use is to read the project files themselves and follow the reference chains. Same logic Clip Sweeper uses on a single drive applies to a NAS, just at scale.

The Pre-Cleanup Announce Protocol

Before any bytes get moved, the team needs to know.

Send the announcement at least one week before. A Slack post in the #post or #editorial channel: "Doing a NAS cleanup pass starting Monday. Anything you have not opened in 18 months will be flagged. If you want a project preserved, mark it." Give people time to respond. Editors travel, take vacation, miss messages.

Freeze writes during the scan. While you are inventorying which projects are active, no one should be saving new versions or moving files around. A two-hour quiet window is usually enough to scan even a large NAS. Schedule it after-hours if writes during business hours are non-negotiable.

Establish a veto. One person on the team (usually the senior editor or post supervisor) has the final call on borderline cases. Without a tiebreaker, every cleanup ends in committee paralysis and nothing gets deleted.

Document the process publicly. Whatever you decide to delete, post the list. Editors should be able to scroll through and say "wait, that is mine, do not touch it" before anything actually moves.

The Phased Cleanup

Always work top to bottom. Each phase has progressively higher risk to active work.

Phase 1: Regen-able files (safe)

Cache, render, and proxy files that the software can rebuild. Zero project risk.

On a typical post-house NAS that is 18 months old, this phase alone reclaims 8 to 15 percent of total capacity. Adobe Media Cache folders, render previews, optimized media, audio peak files, transcoded proxies. Look in Adobe Premiere Pro Auto-Save, Media Cache Files, CFA, PEK, Render Files, Transcoded Media, and any folder named proxy or proxies that sits inside an archived project tree.

The only proxies you preserve in this phase are ones inside currently-active projects. If a project has been delivered and archived, its proxies can be regenerated from camera originals if anyone ever needs them again.

Phase 2: Abandoned tree audit

Project trees with no activity for 18+ months. Higher risk because of misidentification.

Start by listing every project root folder on the NAS. Sort by the most recent modification timestamp inside that folder (not just the folder's own mtime, which is usually meaningless). Anything where the most recent file change is over 18 months old is a candidate.

Before deletion, the senior editor reviews the list. Most post houses find that 60 to 70 percent of NAS contents fall into this bucket on a server that has been running for two years or more without a cleanup. That is not a typo. The majority of a typical post-house NAS is dead weight.

Do not delete in place. Move abandoned trees to a "to-be-deleted" staging folder. Leave them there for 30 days. If nobody complains, delete. If someone needs something back, restore it from staging in two minutes instead of from off-site backup in two days.

Phase 3: Unused media inside active projects

The hardest and highest-value phase. Inside a project that is active, there is usually 30 to 50 percent of media that was imported, considered, and never placed on a timeline.

Removing this media requires reading the project file (or files, if there are linked AE projects) and identifying which media is referenced by a sequence and which is just sitting in a bin. This is exactly what tools like Clip Sweeper do. Doing it manually for a 200-clip project is the same as cross-referencing the script supervisor's notes against every reel in the vault by hand. Hours of work, easy to miss something.

A single deleted clip that was actually used in a Linked After Effects composition can break a delivered final. So this phase always uses cross-project awareness: a file that is unused in Project A but referenced by Project B (or by a linked AE comp) stays on the NAS.

After Cleanup: Keep It Clean

The hardest part of NAS hygiene is preventing the bloat from coming back. A few habits that work.

Naming conventions that survive turnover. Every project folder starts with a year-month prefix: 2026-04-ClientName-ProjectName. The folder structure is uniform across editors. New people on the team can find anything by reading the folder name.

A status flag inside each project root. A simple text file: STATUS.txt containing ACTIVE, DELIVERED, ARCHIVED, or ABANDONED. Editors update it when they finish a job. Cleanup scripts can sort by it. The next person to clean the NAS knows immediately what is what.

Monthly check-in. Twenty minutes at the end of each month. Producer or post supervisor reviews any project that crossed from ACTIVE to DELIVERED and decides whether to archive or keep on the live NAS. A small, regular review prevents the eighteen-month "I cannot face this" cleanup that everyone dreads.

Don't let freelancers leave projects on the NAS forever. Most freelancers say "give me six more months in case the client comes back." Most clients do not come back. Set a 90-day post-delivery retention policy and stick to it. Anything older moves to slower, cheaper archive storage.

For sizing the NAS itself once the bloat is gone, the storage calculator can help you figure out the realistic working footprint and what to size your archive at separately.

Before the Next Cleanup Day

The reason most NAS cleanups are dreaded is that nobody knows what is safe to delete and the manual audit takes weeks. Clip Sweeper reads the actual project XML across every Premiere project on the server and tells you, with cross-project awareness, exactly which media is referenced and which is not. On a typical post-house NAS, that single scan finds tens of terabytes of footage nobody is using. The team gets their server back. Nobody gets paged at midnight because their delivery broke.