Updated: September 10, 2026
The video is delivered, the client is happy, the invoice is paid. Eighteen months later they call: the product changed, can you swap two shots and re-export. The answer depends entirely on decisions someone made the week the project wrapped, when archiving felt like the least urgent thing on the list.
Archiving is not backup. Backup protects work in progress. An archive is a deliberate decision about what a finished project needs to remain useful years from now.
What to keep, and what not to
Keeping everything is expensive and, past a certain point, actively unhelpful, because finding anything in an undifferentiated pile of drives takes longer than reshooting.
Keep the camera original media. The unmodified files from the camera. Everything else can be rebuilt from these; nothing can rebuild them.
Keep production audio. Separate recorder files, not just the camera scratch track.
Keep the project file and its structure. The timeline, bins and settings from the editing application.
Keep graphics and design source files. Layered files and project files for motion graphics, not just the rendered output.
Keep the finished masters. A high quality master export, plus the delivery versions actually supplied.
Keep the paperwork. Releases, licences and the terms that govern how the video may be used. This is the part that gets separated from the media and later cannot be found.
Consider dropping render caches, optimised media, proxies and intermediate exports. All of these can be regenerated and all of them are large.
Whether to keep unused takes is the judgement call. Client footage that could be repurposed usually justifies it; forty takes of a slate does not.
Where it lives
The old rule still holds: more than one copy, on more than one type of media, with one copy in a different physical location.
Spinning drives are cheap and convenient and are not archival media. They fail, and unpowered drives sitting for years fail more often than people assume. Two copies minimum, and a plan to migrate the data before the drives age out.
Cloud storage solves the offsite problem and introduces a cost that continues forever. Archive tiers are cheaper to store and charge to retrieve, which is fine for an archive nobody touches and expensive for one that gets accessed.
LTO tape remains the genuine archival option: long shelf life and low cost per terabyte, at the price of hardware and a workflow.
In Miami there is a further consideration that inland facilities do not share. Hurricane season means an offsite copy should be genuinely offsite, not a second drive in the same building. Humidity also matters for long term storage, so a climate controlled space beats a cupboard.
Naming and documentation
An archive nobody can search is a pile of drives with sentimental value.
Adopt one naming convention and hold it: client, project, date, version. Avoid final, and avoid spaces and special characters that break scripts and cause trouble across operating systems.
Write a plain text index and keep it with each archive and separately. It should say what the project was, when it was shot and delivered, what is on the media, what software and version the project file needs, what fonts and plugins were used, and where the licences and releases are.
That last group is the one that stops a future re-edit dead. A project that needs a plugin nobody has any more, in a version of an application nobody runs, is technically archived and practically lost. Our note on media management before you edit covers the front end of the same discipline.
How long to keep it
There is no universal answer, and the useful approach is to tie retention to something concrete.
Consider how long the video will be in use, whether the client is likely to want versions, what the contract says about retention and who owns the material, and how long any licence for music or stock footage runs. A licence that expires makes the master unusable in its current form regardless of whether you still have the files.
Agree it with the client in writing rather than defaulting to keeping everything forever. If they want long term retention, that is a service with a cost attached, and it is reasonable to say so.
Test the archive
An archive that has never been restored is an assumption.
Verify checksums after writing, so you know the copy is bit for bit identical rather than merely present. Then, periodically, actually restore something: pull a project from a year ago and confirm it opens, relinks and plays.
The failures this catches are mundane and fatal. Media that never made it onto the drive. A project file that relinks to nothing. A codec the current system no longer reads. Finding that out during a test costs an hour; finding it out when a client is waiting costs the relationship.
Editing with a team that plans past delivery
Post-production does not end at export. The projects that are easy to revisit are the ones where somebody spent an hour at wrap organising the media, writing the index and verifying the copies.
When we finish an edit in Miami, the archive is part of the deliverable rather than an afterthought: camera originals, audio, project files, graphics sources, masters and the paperwork, indexed and verified. Clients notice it when they come back two years later and the answer is yes rather than a long pause.
If you have finished projects sitting on unlabelled drives and want them organised into something you can actually use, that is work we do. See our notes on delivery specs for what a master should be, read more on the about page and on the blog, or get in touch.
Frequently asked questions
How long should video projects be archived?
Tie it to the video's useful life, the contract and any licence terms rather than to a fixed number. Agree it with the client in writing so expectations match.
Is cloud storage enough on its own?
It covers the offsite requirement well. Many teams still keep a local copy for fast access and to avoid retrieval charges, which is the multiple media part of the rule.
Should I archive the proxies?
Usually not. Proxies and render caches regenerate from the originals and take up substantial space for no long term benefit.
What if the client wants an edit after the archive is gone?
Then the work starts from whatever remains, often just the master, which limits changes to the level of the finished file. This is exactly why the retention conversation belongs in the contract.