A video asset filing system you can still use in five years
The test is not whether your drive looks tidy today. It is whether you can find the interview you shot two years ago, open the project, and export it again.
8 minute read
The only two tests that matter
A filing system for video has to pass two tests, and both of them happen years after you set it up. The first is whether you can find a specific thing — the second interview from the March shoot, the b-roll of the workshop — without opening files to check what they are. The second is whether the thing you find still opens: the codec still decodes, the project relinks, the drive still spins up. Most people solve neither, because on the day you shoot, neither problem exists. A folder called new folder 2 costs nothing this afternoon and costs you an afternoon in 2029. Everything below is aimed at those two tests. Tidiness for its own sake is not the goal, and a system you will not maintain is worse than a messy one you understand.
One folder shape, repeated for every shoot
File by project, not by file type. A top-level folder per shoot or episode, with an identical set of subfolders inside every one of them, beats a single global Footage folder and a single global Exports folder for one blunt reason: when you archive or delete, you move one object rather than hunting through five. A shape that survives contact with reality is roughly five folders, numbered so they stay in the order you use them — camera originals, audio, project and edit files, graphics and music, exports. Proxies get their own folder, named so it is obvious they are disposable. Resist nesting by day and by camera beyond one level, because deep trees are where files go missing. The point is that every project looks the same, so navigation becomes muscle memory instead of a decision.
Name files so they sort themselves
Stanford Libraries publishes the clearest set of file-naming rules going, and it transfers to video almost unchanged. Lead with a date in YYYYMMDD form so chronological order and alphabetical order become the same thing. Avoid spaces and the punctuation that breaks tools — the guidance names characters like the ampersand, question mark, hash, quotes and slashes as ones to leave out — and use underscores, dashes or camel case as separators instead. Use leading zeros on any sequence, so 001 through 010 sort correctly rather than putting 10 next to 1. Keep names short enough that software does not truncate them, and put the version number in the name rather than relying on modified dates. Stanford's last suggestion is the one everyone skips and the one that saves you: a readme file in the project folder explaining your abbreviations.
Decide what is replaceable before you run out of space
Storage panic makes bad decisions, so make the decisions now, while nothing is full. Camera originals and recorded audio are irreplaceable: the shoot is over, the guest has gone, and no amount of money regenerates them. Project files are tiny and essential, because they are the only record of the decisions you made. Proxies are disposable by definition — they exist to be regenerated from the originals, which is the entire reason editors make them. Exported masters sit in the middle: technically regenerable if the originals and project both survive, but re-exporting a two-year-old timeline is an afternoon you will not want to spend, so keep the finished renders and bin everything between. Anything you can rebuild in under ten minutes is a candidate for deletion. Anything you cannot rebuild at all never gets deleted, only moved somewhere cheaper.
Verify the copy, then keep three of them
A copy that finished without an error message is not a copy you have checked. Offload tools that compute a checksum per file and write a sidecar manifest, such as an ASC MHL, prove the destination matches the card before you format it, and re-running it a year later tells you whether anything has degraded on the shelf. Then keep more than one. The convention worth learning is 3-2-1, which Backblaze states as three copies, on two different devices or media, with one off-site; it notes the United States government has recommended the same, citing a 2012 Carnegie Mellon paper for US-CERT. Drives fail often enough to justify that. Backblaze's Drive Stats for the first quarter of 2026, covering 341,263 drives, put the quarterly annualised failure rate at 1.24% and the lifetime rate at 1.39%, and those are data-centre drives rather than a bus-powered disk in a backpack.
What archive storage actually costs
Cloud storage is not one price, and the gap between tiers is where the decision lives. Amazon's published S3 rates for its US East region run from $0.023 per GB-month on S3 Standard down to $0.00099 on Glacier Deep Archive — roughly $23 versus $0.99 per terabyte per month, a twenty-three-fold difference for the same bytes. The catch is what you give up. Deep Archive retrievals take twelve to forty-eight hours and bill separately, and every archive class has a minimum storage duration you pay for whether or not you keep the file that long: thirty days on Standard-Infrequent Access, ninety on the two mid Glacier tiers, one hundred and eighty on Deep Archive. So archive tiers reward material you are confident you will not touch. Working footage for a project still in progress belongs on the expensive tier, and moving it down early costs more than leaving it.
Will the file still open?
Findability is the problem people plan for. Format obsolescence is the one that actually eats archives, and it arrives silently — a codec that needed a plugin, a camera format the manufacturer stopped supporting, a project file two major versions behind. The Library of Congress maintains a Recommended Formats Statement for exactly this, and for moving image works its preferred tier is uncompressed or lightly compressed material, naming IMF, XDCAM HD or higher and ProRes 422 HQ or higher, with the open lossless codec FFV1 listed as acceptable. You do not need an institutional workflow to take the point. If you are keeping one file forever, keep a widely supported mezzanine format rather than a proprietary camera format, and keep the plain-text transcript beside it. Text outlives every codec, and it is what makes an archive searchable in the first place.
The version of this you will actually do
None of the above requires a weekend. Pick the folder shape and use it on the next shoot only, rather than reorganising the past. Add the date-first naming rule to whatever you export from that day forward. Then do one backwards job: work out which drive holds your irreplaceable camera originals, and get a second copy of that drive somewhere else this week. That single step covers most of the risk. Retro-filing an old archive is satisfying and low value; almost nothing on those drives will be opened again, and the few things that will are usually findable by date. Fix the intake, protect the originals, and let the old mess stay a mess. A system that starts on Monday and survives is worth more than the perfect one you plan for January.
FAQ
How should I name video files?
Date first in YYYYMMDD form, then project or scene, then a short description, then a version number. Skip spaces and punctuation like ampersands, hashes, quotes and slashes, using underscores or dashes instead, and pad any sequence with leading zeros so 001 sorts before 010. Stanford Libraries publishes this as general research-data guidance and it works unchanged for footage.
What should I delete from old video projects?
Proxies, cache and render files, and any intermediate exports. All of those regenerate from the camera originals and the project file, which is why editors treat them as disposable. Keep camera originals, recorded audio, the project file and the final exported masters. The rule of thumb: if you could rebuild it in ten minutes, it does not need archiving.
How long should I keep raw footage?
Longer than you expect to want it, because deleting camera originals is the one irreversible choice in the whole workflow. The practical answer is to move it rather than delete it: Amazon's Glacier Deep Archive lists at $0.00099 per GB-month, about $0.99 per terabyte per month, so a decade of old shoots costs less than a coffee a month to keep. Retrieval takes twelve to forty-eight hours and is billed separately.
Is an external drive or cloud storage better for archiving video?
Both, which is the whole point of the 3-2-1 rule Backblaze describes: three copies, two different devices or media, one of them off-site. A drive gives you fast local access and a cloud archive tier gives you geographic separation for a few dollars per terabyte per month. One drive on a shelf is not a backup, and neither is one cloud bucket.
Sources
- Name files - Data best practices · Stanford Libraries
- The 3-2-1 Backup Strategy · Backblaze
- Backblaze Drive Stats for Q1 2026 · Backblaze
- Amazon S3 pricing · Amazon Web Services
- Recommended Formats Statement · Library of Congress
Related pages
Keep reading
More Video Basics →Social media video specs for every platform
One 1080x1920 export covers most of short-form. This is where that stops being true — the length caps, file limits, and the platforms where the number changes depending on how you upload.
8 minute readSocial media image sizes for every platform
Instagram's help page now allows 3:4, YouTube's recommended thumbnail is 4K, and TikTok publishes almost nothing. Here are the numbers each platform actually documents, and where they leave you guessing.
7 minute readHow to convert landscape video to vertical
A full-height 9:16 crop of a 1920x1080 frame is 607 pixels wide. Which third of the picture you keep, and whether it is allowed to move, is the entire job.
8 minute readTurn one long video into a week of posts
300 credits for 3 days · no card.
Start free