
Somewhere in your office is a drive labeled "2023, SORT LATER." Later never came. RAW file chaos isn't a character flaw; it's what happens when every shoot adds thousands of files and no system decides where they live, what they're called, or when they're allowed to die. Here's a system that makes all three decisions once, so you never make them again.
Everything hangs off a single root folder, not scattered across Desktop, Downloads, and three drives. Inside it, one pattern:
The point isn't this exact scheme. It's that a boring pattern applied identically to every shoot beats a clever one applied when you have energy. Your folder structure should be so predictable that you could find any file from any year while on the phone with a client.
Camera filenames recycle: every few months another _1047.ARW is born, and someday two of them meet in the same folder. Rename on ingest (Lightroom, Photo Mechanic, and most ingest tools do this automatically) to a pattern like 20260602_smith_0001: date, client, sequence. Now every file is globally unique, self-describing, and traceable to its job even if it escapes its folder into an email or a print order.
Two bans worth enforcing: no spaces or special characters (they break scripts, web uploads, and some backup tools), and no version adjectives, final_final_v3_ACTUAL is how deliveries go wrong. Versions belong in your catalog or your Exports folder, not in adjectives.
Here's the math that forces the decision. A full wedding day typically produces 2,000 to 4,000 RAW frames, of which 400 to 800 become delivered images (typical industry ranges, per self-published platform data from Snapeen). That means the majority of every wedding you store, at modern RAW sizes, easily hundreds of gigabytes per job once backups are counted, is frames you already decided weren't good enough. Multiply by a season, then by your 3-2-1 backup copies, and "keep everything" quietly becomes the most expensive subscription you own.
| Policy | What you keep | Storage cost | Risk |
|---|---|---|---|
| Keep everything | Every frame, forever | Highest, compounds yearly | Paying to protect your worst frames |
| Cull-then-archive | Picks + near-picks; hard rejects deleted after delivery | A fraction of everything | Rare "wish I'd kept it" moments |
| Scorched earth | Delivered images only | Minimal | No re-edit or dispute recourse |
The middle path is the sane one for most photographers. Keep your picks and the plausible near-picks (the second-best frame of each scene, cheap insurance for re-edits and close calls). Delete the hard rejects, blinks, focus misses, the lens cap, after the gallery ships and a buffer period passes. Those frames have no future; you rejected them once with fresh eyes and you'll never choose them later. A tight cull, in other words, isn't just a faster edit. It's a permanently smaller storage bill, which is one more reason culling well matters (start with the complete culling guide if yours is ad hoc).
Most photographers keep RAWs forever out of vague guilt: what if they ask? Replace guilt with policy. Choose a retention period you can afford to honor, write it into your contract as an archival clause, "RAW files are retained for X after delivery, then deleted; delivered images are yours to keep", and mention it in the delivery email. (Vague or missing archival terms are a classic contract gap; see the photography contract guide.)
A stated policy does three jobs at once: it lets you reclaim drives on schedule without fear, it turns extended archiving into a sellable add-on rather than a free forever-service, and it sets client expectations before the tearful "we lost our downloads" email arrives, which, with a policy, has a calm answer either way. Where the archive physically lives, working drives vs cold storage vs cloud tiers. Is its own decision, covered in our cloud storage guide.
Every rule above assumes the files are safe first. Nothing gets renamed, culled, or deleted until the cards are copied to at least two places, the 3-2-1 discipline in our backup workflow guide. Deleting rejects from a properly backed-up archive is housekeeping. Deleting anything from your only copy is gambling with someone's wedding.
Hard rejects, blinks, total misfires, lens-cap frames, yes, after the gallery is delivered and a buffer period has passed. They have no future value and they're the bulk of your storage bill. Near-picks are worth archiving; genuinely broken frames are not. The key is making it a written policy you follow, not a monthly agonizing.
Whatever your contract says. That's the real answer. Pick a retention period you can afford to honor, write it into your contract as an archival clause, and tell clients at delivery. A stated policy converts "forever, out of guilt" into a defined service with an end date.
One root photos folder, then year, then a date-plus-client folder (2026-06-02_smith-family), with the same subfolders inside every job: RAW, Selects, Exports, Docs. The structure matters less than the consistency, any system you apply identically to every shoot beats a clever one you apply sometimes.
It's a workflow preference, not a necessity. DNG can embed metadata without sidecar files and offers optional lossy compression for archives; native RAW keeps perfect compatibility with your camera maker's software and avoids a conversion step on ingest. Either is fine. What matters is picking one and being consistent.
Kepla for Mac clears the obvious misses from a card, names the reason on every frame it sets aside, and leaves the choosing to you. Nothing is ever deleted, moved or renamed. Free through the private preview · the first hundred photographers keep it at $99 a year.