Photo archiving guide: a ten year plan

Most photographers have a backup routine and no archive at all. The two are not the same job. A backup is there for the drive that dies on Tuesday. An archive is there for the couple who email you in 2036 asking for a full resolution file from a wedding you barely remember. This is how to build the second one: what goes in it, what format it lives in, where cold copies sit, and how you prove ten years from now that it still opens.

An archive is not a backup

A backup answers a question about this week. A card corrupts on Sunday, a drive dies on Tuesday, you overwrite the wrong folder on Thursday. You reach for the copy, you put it back, you carry on. A backup is a living system. It changes every day, and the whole point is that it mirrors what you have right now.

An archive answers a question nobody asks you until years later. A couple wants a print for a tenth anniversary. A venue wants the full resolution file for a rebrand. A client's house floods and the only surviving copy of their wedding day is the one in your studio. None of that is about Tuesday. It is about whether a set of files you stopped thinking about in 2026 still opens on a computer nobody has designed yet.

Those two jobs pull in opposite directions. A backup wants to be automatic, connected, and constantly in sync. An archive wants to be finished: written once, checked, disconnected, and left alone. That difference matters more than any product choice, because if your only long term copy lives inside the folder your sync software watches, then anything that damages the live copy has a clear path to the copy you were counting on. That includes you, tired, at 1am, tidying up.

This guide assumes the weekly side is already handled. If it is not, start with our backup workflow for photographers and come back. Everything below picks up where that stops.

Decide what goes in the archive before you decide where it lives

People buy the drive first and then argue about the policy forever. Do it the other way round. A wedding day typically produces 2,000 to 4,000 RAW frames and around 400 to 800 of them get edited and delivered. That gap is the entire archiving question. Everything else is shopping.

There are three tiers worth naming, and they do not need the same treatment:

• Delivered files. The finished gallery, at full resolution, exactly as the client received it. This is the tier you will be asked for. It is also the smallest, so there is no excuse for it not being in every copy you own.
• Picked originals plus edit settings. The RAW frames behind the gallery, with their sidecars or catalog settings. This is what lets you re-export at a different size, fix a crop, or rebuild a lost gallery.
• Everything else from the shoot. The frames that did not make the gallery. Keeping these is a real decision with a real cost, and we work through it properly in what to do with the photos you don't pick. The short version: the cost per booking is usually small, and the requests that arrive years later are often for a person rather than a better frame.

Write your answer down as one sentence per tier, with a retention period attached, and put it in the same document as your delivery standards. If you are unsure what a normal gallery size looks like before you set the policy, our piece on how many wedding photos to deliver has the ranges. A policy you can say out loud beats a hard drive you keep meaning to sort.

Formats: pick what will still open in 2036

An archive of proprietary RAW files is a quiet bet that your camera manufacturer will still care about a discontinued body in fifteen years. Sometimes that bet pays. Sometimes a format from a company that left the camera business becomes a hunt for old software on a machine that can still run it.

The mainstream hedge is DNG. Adobe publishes the DNG specification publicly, grants a royalty free right to build software that reads and writes it, and offers a free converter. Adobe's own framing is that an open specification helps archival confidence, and it is one of the few claims in this space you can check yourself by reading the license. That does not make DNG compulsory. It makes it the format with the fewest single points of failure.

A workable rule for each archived shoot:

• Originals: keep the camera's own RAW files, and if you want the hedge, keep a DNG conversion alongside rather than instead.
• Deliverables: full resolution JPEG, which every device on earth reads, and TIFF for anything you would be upset to re-edit.
• Edit settings: sidecar files written next to the images, not settings that only exist inside one catalog on one computer.

That last point is the one that catches people. A catalog is software, not an archive. If the only record of your edits is a database file tied to one application version, you have made your work dependent on that application still installing a decade from now. Metadata written into the files themselves travels much better, and our guide to photo metadata covers what is worth embedding.

Cold storage, in plain terms

Cold storage just means a copy that is not plugged in, not syncing, and not waiting for a command from your computer. Its resilience comes mostly from being disconnected. The old rule still holds: three copies of anything you care about, on two different kinds of media, with one of them somewhere else entirely.

Prices for all of this move constantly, so work them out for your own volume rather than trusting a number in an article. Measure one of your own RAW files, multiply by your unpicked frames per shoot, multiply by shoots per year, and you have your annual growth. Then compare that against drives or against your provider's next tier, and against what one shoot earns. Our breakdown of cloud storage for photographers goes through the options in detail.

Make each shoot explain itself

The person who has to understand your archive is you, ten years older, with no memory of the job and no access to the software you were using at the time. Possibly it is not even you. An archive that only makes sense to someone with the catalog open is an archive with a dependency.

Two habits fix almost all of it. First, folder and file names that carry meaning on their own: date first so things sort chronologically, then client, then shoot type. Anything you can read from the filename is something you never have to open an application to find. We lay out a full scheme in our photo file naming system.

Second, a plain text file inside each archived shoot folder. Not a document, not a spreadsheet, just a text file that any computer since 1980 can open. Five lines is enough:

• Who the clients were and when the shoot happened.
• What is in each subfolder, in one sentence each.
• Which software wrote the edit settings, and which version.
• How many files were in the shoot when it was sealed, and the total size.
• The date the archive copy was made and verified.

That fourth line does more work than it looks like. A count and a size are how you notice later that something is missing, which is the failure mode nobody plans for: not a drive dying loudly, but a folder that quietly came back short from a copy three years ago.

The part everyone skips: proving it still reads

An archive nobody has opened is not an archive, it is a hope. Files decay in undramatic ways. A copy interrupted halfway through, a drive with a handful of bad sectors, a cloud sync that hit a permissions error at 2am and gave up on one folder. None of it announces itself. You find out on the day someone asks for the files, which is the worst possible day to find out.

Two practices cover it, and both are boring by design.

Checksums. When you seal a shoot, generate a manifest of file hashes and store it in the folder with the images. Most backup and ingest tools will do this for you, and both macOS and Windows can produce hashes from the command line without installing anything. A year later you re-run the check and compare. Identical means the bits are intact. A mismatch means you replace that file from another copy today, while another copy still exists.

Restore drills. Once or twice a year, pick a random shoot from a random archive medium and actually restore it. Open a RAW file. Open a delivered JPEG. Read the text file. Time how long the whole thing took. You are testing three things at once: that the media reads, that the software still opens the format, and that you still remember how your own system works. If a drill fails, you have found the problem while it is a chore rather than an emergency. If it fails badly enough that files are gone, our guide to recovering deleted photos covers what to do next and, more importantly, what not to do first.

A ten year plan you can actually keep

Nothing here survives a decade because the media is good. It survives because you kept copying it forward onto whatever is current before the old thing became unreadable. Migration is the whole strategy. Everything else is housekeeping.

One more thing shapes how big all of this gets: what your selection process does to the frames that did not make the gallery. If your workflow physically discards them, your archive is smaller and permanently missing the frames a client might ask for years later. If it only marks which photos were picked and leaves every original in place, your archive is complete and the decision stays reversible. That is the approach we are building Kepla around: it picks your best shots and hands you a shortlist, and it never deletes, moves or renames a single file. The picking app for iPhone, iPad and Mac is still in development. The booking page is live and free today while we build the rest.

FAQ