How to Organize Old Family Photos (Before You Scan)
The step most guides skip: identify the faces while someone can still name them. A four-stage system from the Library of Congress, adapted for a shoebox.
How to Organize Old Family Photos (Before You Scan)
Almost every guide to organizing family photos starts with "gather everything in one place." That's fine advice, and it's not the first step. The first step is finding out who the people are, because that information lives in a small number of aging heads and it's the only part of this project with a deadline.
Everything else — sorting, scanning, restoring, backing up — can be done by anyone, badly, in ten years, and still be recoverable. A box of unlabeled faces is not.
Stage 1: Identify, while you can
Before you sort a single stack, pick the twenty or thirty photos with unidentified people and sit down with whoever in your family is oldest and sharpest. Record the conversation on your phone — don't take notes, you'll miss things while writing.
Ask about the specific and the incidental: who is that, where was this, whose house, what was the occasion, who took the picture, why is he wearing that. The stories attach to the pictures and they're most of the value. You can transcribe later; you cannot re-run the conversation later.
Write directly on the back of prints only with a soft pencil, and only in the margin — never a ballpoint or marker, which press through and bleed over time. If that makes you nervous, put the print in a sleeve with a slip of paper instead.
Stage 2: Decide what's actually in scope
This is where projects die. A shoebox is finite; four boxes and three albums is a hobby you'll abandon in April.
The Library of Congress's personal archiving guidance for photographs puts it plainly at the second step: "Pick the images you feel are especially important. You can pick a few photos or many. If there are multiple versions of an important photo, save the one with highest quality."
Move fast and shallow rather than evaluating each photo:
- Go through a stack looking only for keepers, not judging every frame.
- Near-duplicates: keep the sharpest one. Nobody has ever wanted the other four shots of the same porch.
- Unidentifiable landscapes and buildings with no people: a small representative handful, not all sixty.
- Anything with a person you can name, an event, or a room you recognize: keep.
- Damaged photos of people who matter: keep, obviously. Damage is fixable; a missing photo isn't.
Aim to cut the pile by half or more. You are not deleting family history; you are choosing what gets the expensive treatment.
The parking-lot pile
Keep a third pile next to keep and discard: "ask someone." Anything you can't decide in three seconds goes there and stops blocking you. Most people find the ask-someone pile answers itself once the rest is sorted into a shape.
Stage 3: Sort into a structure you can name out loud
Pick one organizing axis and stick to it. The three that work:
- By family line — best if you're heading toward genealogy, and it maps onto how relatives think.
- By decade — best when identification is patchy, because you can usually date a photo from clothes, cars, and print borders even when you can't name the people.
- By household or event — best for recent decades with lots of birthdays and holidays.
Mixing axes is what produces a system only you understand. If you can't say your structure in one sentence — "top level is family surname, second level is decade" — it's too complicated.
Then do the same thing digitally. LOC's third step is to "create a directory/folder structure on your computer" and to "write a brief description of the directory structure and the photos." That last instruction is the one nobody follows and the one that matters most: a plain README.txt at the top of the folder saying what the structure is, what the naming pattern means, and who the abbreviations refer to. You are writing it for a nephew in 2060 who has no idea what RS stands for.
For file names, LOC asks for descriptive names plus tagging with names of people and subjects. A pattern that survives every operating system and sorting quirk:
1974-08_smith-reunion_grandma-ruth_01.tif
1958_unknown-wedding_maybe-aunt-cecile_03.tif
Date first so chronological order is free. Underscores between fields, hyphens inside them. And note the second example: "unknown" and "maybe" are legitimate metadata. Recording your uncertainty is far more useful to a future reader than a confident guess that turns out wrong.
Stage 4: Digitize, then repair — in that order
Only now does scanning make sense, because you're scanning a curated set instead of a shoebox. Settings, DPI, and formats are their own subject — how to scan old photos covers them, and slides and negatives need a different rig entirely, covered in digitizing slides and negatives.
Repair comes after scanning, never before, and only on the photos that earn it. Running photo restoration on 400 images because it's cheap gives you 400 files you now have to re-sort. Restore the ones going on a wall, into a book, or into a family group chat.
One filing rule that saves real pain: keep the restored version alongside the original scan, never instead of it. Same folder, same name, one suffix:
1974-08_smith-reunion_grandma-ruth_01.tif
1974-08_smith-reunion_grandma-ruth_01_restored.tif
Restoration is an interpretation. If a later tool does better, or if someone objects that Grandma Ruth's eyes came out wrong, you want the untouched scan still sitting there. The full restoration workflow covers what order to run things in.
Stage 5: Copies, and the part everyone skips
LOC's last step is the most concrete guidance in the whole document, and it's stricter than what most family archives actually do:
- Make at least two copies — more is better. One can live on your computer; put others on separate media.
- Store copies in locations "as physically far apart as practical." A second drive in the same house is not a backup; it's the same fire.
- Put a copy of the photo inventory with your important papers in a secure location.
- Check your photos at least once a year to make sure you can still read them.
- Create new media copies every five years, or sooner if needed, to avoid data loss.
That five-year replacement cycle is the number nobody tells you. Drives and discs are consumables, not archives. Put a recurring calendar reminder in now, while you care about this project — future you will not spontaneously remember in 2031.
Physical storage of the prints themselves — temperature, humidity, enclosures, what "archival" actually means — is a separate subject with real specifications, and it's in how to preserve old photos.
The order, in one list
- Record the stories from whoever can still tell them.
- Cut the pile in half; keep the best copy of duplicates.
- Sort on one axis, and write down what the axis is.
- Scan the survivors; name files date-first; keep a README.
- Restore selectively, keeping originals beside restored copies.
- Two copies minimum, in two places; check yearly; new media every five years.
Do them out of order and you'll do most of them twice.
FAQ
What's the first step? Identification. Sit with the oldest relative who can name faces and record it. Everything else can wait; that can't.
Should I scan everything first? No. Sort first — scanning is the slow step, and scanning duplicates is how these projects stall.
How should I name the files? Descriptively, date-first, with people's names in them: 1974-08_smith-reunion_grandma-ruth_01.tif. Record uncertainty with "unknown" or "maybe" rather than guessing.
How many backups do I need? The Library of Congress says at least two copies, stored as far apart as practical, checked yearly, with new media copies made every five years.