Most editors finish a project, hit deliver, and then leave the whole thing sitting on the same Frame.io or LucidLink workspace where they cut it. Six months later, that "wrap" is still costing $240 a month. Two years in, you've paid more in rent than the camera owner paid for the lenses.
There is a better way. It isn't complicated, but almost no one does it because nobody handed them a checklist on day one. Here is the checklist.
The Three Places a Finished Project Can Live
Think of finished projects like film prints. The studio kept one in the editorial vault for revisions, one at the lab for distribution, and one in cold storage in a salt mine in Kansas. Each had a different job, and each cost a different amount.
Your modern equivalent is three tiers:
| Tier | Best for | Typical cost (20 TB, 5 years) |
|---|---|---|
| Live cloud (Frame.io, LucidLink) | Active project, daily access | $14,400 |
| Cold cloud (Backblaze B2, Wasabi) | Recently finished, occasional pulls | $8,400 |
| Physical archive (LTO-9, cold HDD) | Deep archive, "in case" | $1,200 one-time |
These aren't competing options. They're sequential ones. A healthy project lives on tier 1 while it's active, moves to tier 2 once it's delivered, and lands on tier 3 once a year passes without anyone touching it.
The 5-Year Math, Spelled Out
Take a single 20 TB wedding-and-events catalog. Five years of storage, no growth.
On Frame.io Storage at the going rate, you're looking at roughly $240 a month, or $14,400 over five years. That includes all the bells: review tools, share links, V2 Drive desktop access. You will not use most of those features once the deliverable is signed off.
On Backblaze B2 at $6.95/TB, the same 20 TB runs about $139 a month. Five years works out to around $8,400. You give up the live workflow, but you can still pull files in minutes when a client comes back asking for the unedited reel.
On LTO-9 tape, you buy a drive once (around $4,000 used, $5,500 new) and a few tapes (about $80 each at 18 TB native). Two tapes cover 36 TB with room to spare. Total bill: roughly $1,200 if you already have the drive amortized across other projects, or $5,200 if this is your first tape job.
The cliff between tier 2 and tier 3 is where the savings stop being interesting and start being shocking. A studio sitting on 100 TB of finished work pays $42,000 a decade for B2 versus a one-time $7,000 to bring it all in-house on tape.
The Hybrid Workflow That Actually Works
Here is the pattern that works for almost every kind of editor, scaled to whatever your output looks like.
Days 0 to 90 after delivery: leave it on the live cloud. Clients revise. Producers ask for re-cuts. Color comes back with notes. You'll touch the project enough that paying live rent is fine for a quarter.
Day 90 to 365: move it to B2 or Wasabi. The active edit is over, but you might still get a pull request or a "can you re-export at 9:16" email. B2 retrieval is fast enough that you can answer in the same business day. You're paying a third of what you were paying on Frame.io.
Year 1+: move to LTO or a cold HDD shelf. This is the real archive. The clip you might need in 2031 when the client wants to license a five-second moment for a documentary. It costs essentially nothing to keep, but pulling it back will take you a day with a tape mount and an LTFS browse.
This is the same shape as the 3-2-1 backup rule most IT people will lecture you about, except adapted for the reality that video files are huge and your time isn't free.
Strip the Project Before You Archive
Here's the part that almost nobody does, and the reason most archives are double or triple the size they should be.
Before you push a finished project to cold storage, strip it. Cache files don't need to come along. Proxies don't need to come along. And the 30 to 60 percent of imported media that never made it onto a single timeline definitely doesn't need to come along.
A typical 20 TB project that's been "wrapped" still contains:
- 10 to 12 TB of camera originals (some used, some never touched)
- 4 to 6 TB of proxies and transcodes
- 1 to 2 TB of cache
- 1 to 2 TB of versioned exports
A lean archive is just the camera originals you actually used, the project file, and one master export per deliverable. That cuts your archive size in half before you even pick a destination.
The script supervisor doesn't keep the take where the boom dipped into frame. Neither should your archive.
What Lean Archive Looks Like in Practice
For a finished commercial:
- Camera originals from clips that touched any timeline (not the bin, the timeline)
- The final approved .prproj
- One master mezzanine export (usually ProRes 422 HQ)
- One web-spec H.264 of the final cut
- A short README with the spot title, client, delivery date, and any rights notes
For a documentary or feature, add the audio session and any color grade XMLs.
For a wedding, add the full ceremony and toast camera angles even if you didn't use them. Couples come back for re-edits more than any other client type, and you'll regret tossing the wide on the toast.
A Note on Stripping Without Breaking Things
The danger of "strip the project before archive" is that you grab the wrong files and accidentally delete something you needed. This is exactly the problem Clip Sweeper was built for. It reads the actual edit, follows the SubClip-to-MasterClip-to-Media chain Premiere uses internally, and tells you which files are touching a timeline and which were imported and forgotten. Before you push 20 TB to LTO, it's worth knowing that maybe only 12 of those terabytes are actually load-bearing.
A cleaner archive costs less to keep and less to retrieve. It also means you're not paying to store the takes nobody used.