Tagging images is the difference between a photo library you can search and a pile of files you can only scroll. A tag is just a keyword you attach to a picture — "beach," "grandma," "invoice," "2019 road trip" — so that months later you can find every shot of a person, place, or subject in seconds instead of clicking through thousands of thumbnails. This guide explains what a photo tag actually is, the practical ways to tag images on Windows, the mistakes that make tagging a waste of time, and how AI can now do most of the work locally without ever uploading your photos.
One quick scope note: this article is about tagging images to organize your own library, not about adding SEO keywords or alt text to pictures on a website. Those share the word "tag," but they are different jobs.
What is a photo tag?
A photo tag is a short label attached to an image so you can group and find it later. If a filename answers "which file is this," a tag answers "what is this a picture of." One photo can carry many tags at once — a single frame might be tagged wedding, outdoor, Sarah, and 2023 — and you can then filter your library by any combination of them.
There is one distinction worth understanding before you start, because it decides whether your tags survive:
- Embedded metadata tags. These are written into the image file itself, in a standard field (IPTC or XMP "keywords"). Because they live inside the file, they travel with the photo when you copy it to another drive or open it in a different app.
- Catalog tags. These are stored in a photo app's own database, next to the file rather than inside it. They are fast and flexible, but they belong to that app — move the photo somewhere else and the tag may not come along unless the app writes it back to the file.
Neither is "right." The point is to know which kind you are creating, so you are not surprised later when tags do or don't follow your photos around.
Manual tags vs automatic tags
There are two ways tags get onto your photos, and a good workflow uses both.
Manual tagging
You type the tags yourself. This is precise — you know exactly what Sarah or Q3-clients means — but it does not scale. Hand-tagging a few hundred photos is fine; hand-tagging 40,000 is a project no one finishes.
Automatic (AI) tagging
Software looks at each picture and suggests keywords — "dog," "sunset," "mountain" — and often writes a short caption too. AI tagging is what makes a large library searchable in an afternoon instead of a month. The trade-off is that automatic keywords are broad and occasionally wrong, so the best results come from letting AI do the bulk pass and adding your own specific tags (names, projects, events) on top.
How to tag images on Windows
Here are the common methods, from most manual to most automated.
1. Windows File Explorer
Right-click a photo, open Properties → Details, and you'll find a Tags field you can type into; the Details pane in File Explorer shows the same field. It is built in and free, and it writes tags into the file for common formats like JPEG. The limits are real, though: it does not work for PNG or for RAW files, editing tags on many photos at once is clumsy, and there is no way to auto-suggest anything.
2. A metadata / keyword editor
Dedicated metadata tools let you write IPTC or XMP keywords straight into the file, which is the portable option — those tags will show up in any app that reads standard metadata. This is the right approach if you care most about your tags outliving any single program. The downside is that it is still manual, and the interfaces are often utilitarian.
3. A photo manager with AI tagging
For a real library, a photo manager that can auto-tag and let you filter by tag is the biggest time-saver. When you choose one, look for: automatic keywording so you are not typing everything by hand, support for the RAW and HEIC files your camera actually shoots, the ability to filter and combine tags quickly, and — if your photos are personal — processing that stays on your own computer instead of uploading your library to a server to analyze it.
Common tagging mistakes
These are the habits that turn tagging into wasted effort:
- Inconsistent vocabulary. Tagging some photos dog, others dogs, and others puppy splits one subject across three tags. Pick a word and stick to it.
- Over-tagging. Twenty tags on every photo is as useless as none — nothing stands out. Tag the things you would actually search for.
- Tagging in a silo. If your tags live only inside one app's catalog and never get written back to the files, you can lose all of them by switching tools.
- Relying on folders instead of tags. A photo can only live in one folder, but it can carry many tags. Folders and tags solve different problems; tags are what let one image belong to "2023," "Sarah," and "beach" at the same time.
- Not backing up first. Before you run any bulk tagging or metadata-writing pass across thousands of files, make sure the library is backed up.
How Memora tags images locally
Memora is a local-first photo manager for Windows that tags images two ways at once. During indexing it runs a local AI model that writes a short caption for each photo and adds descriptive keywords, and it stores your own tags alongside them so you can filter your whole library by tag. The AI captioning and keywording happen automatically as your photos are indexed; managing your tags and filtering by tag are part of the Pro tier.
The part that sets it apart is where all of this runs. Every step — captioning, keywording, and search — happens on your own PC. No photo, caption, or tag is ever uploaded, so a private library of family or client work stays private, and tagging works offline once the app is set up. That is the honest contrast with cloud photo services, which analyze your pictures on their servers.
Two honest caveats. First, automatic captions and keywords are helpful but can be generic or occasionally off, and captioning runs on the CPU, so it is slower than search — treat the AI pass as a strong first draft you refine with your own specific tags. Second, Memora keeps your tags in its own local catalog; it is a Windows-only app, and tagging applies to photos, not videos. Because tag management sits in the Pro tier, check the current pricing page for what is included; the Free tier lets you browse and run basic AI search first so you can see how it feels.
Tagging vs just searching
Here is the shortcut most people miss: with local AI search, you don't always have to tag every photo to find it. Memora's AI semantic search lets you describe what you want — "dog on a beach" — and surfaces matching photos even if you never tagged them, because the caption and visual model already understand the picture. Manual tags are still worth adding for the things AI can't guess — a person's name, a client, a project code — but they no longer have to carry the whole load. If you want to dig into how those two approaches compare, see image keywording vs AI search.
A simple workflow to tag your library
- Back up first. Confirm your photos exist on a second drive before any bulk tagging pass.
- Let AI do the broad pass. Auto-keyword and caption the library so every photo has a searchable starting point.
- Add the tags AI can't guess. Names, events, clients, project codes — the specifics that matter to you.
- Keep your vocabulary consistent. Settle on one word per subject (dog, not dog/dogs/puppy) so filters stay clean.
- Filter, don't scroll. Find photos by combining tags and search instead of hunting through folders.
Quick checklist before you tag
- Is the library backed up right now?
- Do you know whether your tags are being written into the files or only into an app's catalog?
- Are you using one consistent word per subject?
- Are you letting AI handle the broad keywords and reserving manual tags for names and specifics?
- Is the tagging running locally, or uploading your photos to be analyzed?
Answer those five questions and tagging images stops being a chore you dread and becomes the thing that finally makes your photo library searchable.