Wedding Footage Backup and Archiving: What to Keep, for How Long, and What It Costs

Search for a wedding backup workflow and almost everything you find was written by a photographer. The advice is good advice, for photographs. A full wedding gallery is somewhere around sixty to eighty gigabytes. A two camera wedding day shot in 4K, with a second body, drone clips, and separate audio recorders, is commonly four hundred gigabytes and can pass a terabyte.

At that scale, several rules of thumb that work perfectly for stills quietly stop working. This is the version written for video, including the one setting that gives more filmmakers a false sense of security than anything else in the workflow.

Why photo backup advice does not transfer

Three differences matter, and the first one is the expensive one.

Relay mode is not a backup

The standard photography advice is to shoot to dual card slots so every frame lands on two cards at once. Correct, and for stills it is nearly free. For video it is more complicated in a way that catches people out.

Dual slot cameras generally offer two different modes. Simultaneous, sometimes called backup, writes the same footage to both cards at once. Relay fills the first card and then automatically continues onto the second when it runs out. Relay is a capacity feature. It produces exactly one copy of every clip. A great many filmmakers are shooting weddings in relay mode believing that having two cards in the camera means they are protected, and they are not.

Simultaneous mode has its own catch. The camera writes at the speed of the slower card, and both cards must independently sustain your full bitrate. At high bitrate codecs that means fast cards in both slots, not a fast card in slot one and whatever was in the bag in slot two. If the second card cannot keep up, you are choosing between dropped frames and no backup.

Check which mode your bodies are actually in before the next wedding. It takes a minute per camera and it is the single highest return thing in this article.

Volume changes what is affordable

Keeping five copies of eighty gigabytes is trivial. Keeping five copies of eight hundred gigabytes, times twenty five weddings a year, is a capital expense. Every decision below is really a decision about what you can afford to keep, which is why the pruning section matters more than the storage vendor section.

There is no catalog holding it together

Photographers have Lightroom keeping track of where everything is. A video archive is folders on drives. If the naming is inconsistent, or the drive labels are in your head rather than written down, you have a pile rather than an archive. Two years later, that difference is everything.

The offload is the one place being slow is correct

Almost every permanent loss happens in the twenty four hours between the wedding ending and the footage being safely in two places. The discipline is boring and it is the whole game.

• Never format a card in the field. Not to free up space, not because you are sure you already copied it. Carry more cards instead. Cards are cheaper than a lost ceremony.
• Copy, do not move. A move operation that fails partway can leave you with neither a complete source nor a complete destination.
• Verify with checksums, not with file counts. This is the step photographers can usually skip and video cannot. A partially written large file is still a file, still shows a plausible size, and will not reveal itself until someone tries to play the middle of it. Purpose built offload tools do this automatically, and rsync or robocopy with verification will do it for free.
• Two verified copies before any card is reused. Three before you consider the day closed, with at least one of them somewhere the first two are not.
• Keep the cards themselves as the third copy until the film is delivered, if your card budget allows it. It is the cheapest offsite copy in the business.

The rule worth internalizing: a card is not empty until its contents exist in two verified places, and one of those places is not in the same bag as the other.

When a card does fail, the first minute decides the outcome

Search for this and you will get page after page from data recovery companies. Their central advice is genuinely correct, and worth noticing that it is also free: stop using the card immediately. What those pages tend not to cover is everything around that moment.

• Stop. Do not shoot another frame on the card, and do not let the camera run its own repair or format prompt. Both can overwrite the data you are trying to save.
• Do not run three different recovery apps in sequence. Each one writes to the card and each attempt after the first is worse than the one before.
• Make a sector by sector image of the card first, then run recovery attempts against the copy, never the original. If the first attempt makes things worse you can start over from the image.
• DIY recovery software is reasonable for a card that mounts and shows missing or corrupted files. It is the wrong tool for a card with physical damage, or one that does not mount at all. At that point every attempt reduces what a professional lab could have recovered.

Then there is the part nobody writes about, which is the conversation. If footage is genuinely gone, tell the couple early, tell them precisely what is affected, and arrive with a remedy already decided rather than an apology. Delay is what turns a recoverable relationship into a review and a legal letter. Being the person who called on Monday rather than the person who went quiet until the delivery deadline is most of the difference.

Backup and archive are two different jobs

Conflating these is why people run out of fast storage every September.

• Backup protects work in progress. It is fast, it is written and rewritten constantly, and it only needs to cover the weeks between the wedding and delivery. Speed matters, cost per terabyte matters less.
• Archive protects finished work. It is cold, almost never read, and it needs to survive for years. Cost per terabyte is everything, speed is close to irrelevant.

Using expensive fast storage as your archive is the most common and most costly version of this mistake. Working drives fill with three seasons of finished weddings, and the fix always arrives as an emergency purchase in peak season.

What to archive is the decision that saves the money

Before comparing storage prices, decide what is actually going into storage. There are three realistic tiers.

• Everything. Every card, every angle, every unused take. Simple, safest, and by a wide margin the most expensive.
• Selects, project file, and finished deliverables. The clips that made the cut or nearly did, the timeline, and the exports. Typically a small fraction of the original volume, and it covers essentially every realistic future request: a re-edit, an anniversary cut, a social version, a correction.
• Deliverables only. The finished films and nothing else. Tiny and cheap, but you can never re-cut, so a request for a longer version means telling the couple no.

For most studios the middle tier is the right answer, and choosing it is what makes the arithmetic below survivable. Pruning is a bigger lever than any storage vendor.

The cost math almost nobody actually runs

Take a filmmaker shooting twenty five weddings a year at an average of five hundred gigabytes of raw footage. That is twelve and a half terabytes a year if nothing is pruned.

Cloud archive pricing at the time of writing sits in two broad camps. General purpose storage such as Backblaze B2 is around six dollars per terabyte per month with no retrieval charge. Deep cold storage such as AWS Glacier Deep Archive is roughly a dollar per terabyte per month, but retrieval runs about twenty dollars per terabyte and takes hours rather than seconds. Check current rates before committing, because these move.

Run one year of unpruned footage through both. Twelve and a half terabytes at six dollars is about seventy five dollars a month, or nine hundred a year. The same data in deep cold storage is about twelve dollars a month, but pulling a full year back costs around two hundred and fifty dollars and a long wait.

Here is the part that catches people. Archive cost is cumulative. In year five you are not paying for twelve and a half terabytes, you are paying for sixty two. That is roughly four hundred and fifty dollars a month on general purpose storage, for footage you have not opened since it was delivered. Model the five year number, never the first year.

Now apply pruning. If selects and project files come to a fifth of the raw volume, year five is twelve and a half terabytes instead of sixty two, and the same decision that made storage affordable also made the archive searchable. Local drives change the shape of the numbers but not this conclusion: two copies of every drive, plus replacement every few years, plus the cost of the shelf space and your time managing it.

How long to keep it, and what to promise

There is no industry standard here. Practice ranges from ninety days to indefinitely, and most studios land somewhere between one and three years without ever writing it down.

Write it down. An undocumented retention policy defaults to whatever you happen to have kept, which means the answer to a couple asking for a re-edit is a search through a shelf of unlabeled drives.

A defensible default: deliverables kept indefinitely, because they are small and they are the thing people actually ask for. Selects and project files for two to three years, which covers the anniversary re-edit window. Full raw for a defined period after delivery, then deleted on schedule.

Do not promise forever in a contract. It is an unbounded liability with a compounding cost, and it is a promise that will eventually be tested by a drive failure rather than by your good intentions. State the retention period in writing, state that the delivered film is the couple's to keep, and tell them plainly to download and back up their own copy. You are their videographer, not their backup service.

The question outsourcing raises: who is holding the third copy?

The moment you upload four hundred gigabytes to an editing partner, a complete copy of a couple's wedding day lives on infrastructure you do not control. This is worth thinking about clearly rather than not at all, and it cuts both ways.

The upside is real. For the weeks your partner holds the footage, you have a genuine offsite copy maintained by someone whose business depends on not losing it. During the edit window that is meaningful redundancy.

The trap is treating it as one of your two required copies. It is temporary by design. When the project closes, that copy is on a deletion schedule, and if your local drive fails the following spring the partner's copy may already be gone.

Settle these in writing before the season, not after an incident:

• How long is footage retained after final delivery, and is deletion on a schedule or on request?
• Is it encrypted at rest, and who on their side can access it?
• Is the project file retained as well as the media, and can it be re-supplied later if you need to re-open the edit?
• What is the process if you need something restored, and how long does it take?

Sending organized, clearly labeled media makes all of this easier to reason about, and it is the same discipline that gets you a better first cut.

A policy you can actually run

• Confirm every camera is in simultaneous mode, not relay, with cards fast enough in both slots.
• Two verified copies before any card is formatted. Checksums, not file counts.
• Keep working files on fast storage, finished work on cheap cold storage, and never let the two blur.
• Prune to selects, project, and deliverables once the film is signed off.
• Write your retention periods down, put them in the client contract, and tell couples to keep their own copy.
• Agree retention and restoration terms with your editing partner in writing, and never count their copy as one of yours.
• Model the five year archive cost, not the first year.

The bottom line

Nearly every catastrophic loss in this industry traces back to one of two things: a card formatted before the footage was verified somewhere else, or a single copy that was quietly single the whole time because relay mode looked like redundancy. Both are free to fix and neither requires new equipment.

The archive question is different. That one is a real recurring cost, it compounds every year you stay in business, and the only lever that meaningfully controls it is deciding what not to keep. Make that decision deliberately, once, and write it down. The alternative is making it accidentally, every September, when the drives fill up.