The short version: a category is where an asset lives — its place in a hierarchy. A keyword is what it’s about, ideally chosen from a controlled list so everyone uses the same words. A tag is a free-form label anyone can invent on the spot. You don’t choose one; a well-run library uses all three, each for the job it’s good at. The confusion is real, though — tools genuinely use these words to mean different things, which is most of why people get stuck. The tools that make this structure easy to maintain are compared in our photo library management ranking.
The short answer, side by side
| Category | Keyword | Tag | |
|---|---|---|---|
| Answers | Where does it live? | What is it about? | Any free label |
| Structure | Hierarchy (tree) | Controlled list (may nest) | Flat, uncontrolled |
| How many per asset | Usually one main home | Many | Many |
| Example | Campaigns > Spring 2026 | beach, model-release, CMYK | #reshoot, #fav |
| Best for | Browsing & structure | Consistent, findable search | Quick, personal, temporary notes |
Categories — where an asset lives
Categories are the hierarchy: broad at the top, specific as you go down, with each asset filed into a primary home. Think Campaigns > Spring 2026 > Hero shots. Their strength is browsing and mental structure — a new team member can open the tree and understand how the library is organised without knowing any search terms. That’s something a flat pile of keywords can never give you.
Their weakness is the flip side: a thing can only sensibly live in one main place, so categories are bad at anything that cuts across the structure. A photo of a red convertible at a beach shoot for the spring campaign could belong under vehicles, under locations, or under campaigns — and if you try to solve that by nesting ever-deeper folders, you get a tree so complex nobody files correctly. The rule of thumb: use categories for the one structural home, and let keywords handle everything that cross-cuts. Our guide to organizing a photo library goes deeper on building the tree.
Keywords — what an asset is about
Keywords describe content, and unlike categories you apply many of them to a single asset: beach, convertible, red, model-release, CMYK can all sit on one photo. That’s what makes them powerful for search — you find the asset from any angle, not just the one folder it lives in.
The thing that separates keywords that work from keywords that don’t is control. A controlled vocabulary is an approved list with one preferred term per concept and synonyms mapped to it, so “car,” “cars” and “automobile” resolve to a single searchable keyword. The information-standards body NISO, in the US standard for controlled vocabularies, frames the whole point precisely: natural language has many words for one concept and one spelling for many concepts, so a vocabulary exists “to achieve consistency in the description of content objects and to facilitate retrieval.” Keywords can also nest — a parent keyword animals containing dogs and cats — which is exactly how Lightroom and Adobe Bridge model them. Drawn from a controlled list, keywords are the single most useful layer in a searchable library. Left uncontrolled, they degrade into the next category down.
Tags — the free-form layer
A tag, strictly, is a keyword with no vocabulary control — a label anyone can invent on the spot. The term comes from the idea of a folksonomy, coined in 2004 for “the result of personal free tagging of information and objects… for one’s own retrieval.” Tags are genuinely useful for fast, personal, temporary notes: #reshoot, #favourite, #send-to-client — the ephemeral stuff that doesn’t deserve a place in your formal vocabulary.
The danger is tag sprawl. Because tags are uncontrolled, a team of ten will produce beach, Beach, beaches, seaside and coast for the same idea within a month, and a library with four thousand near-duplicate tags is no more findable than one with none. The fix isn’t to ban tags — it’s to keep them in their lane: use free tags for the throwaway and the personal, and promote anything that turns out to matter into a proper controlled keyword. When a tag keeps getting reused, that’s the signal it has earned a place in the vocabulary.
Open any library that’s been running a few years with free tags switched on and you’ll find the tag graveyard: holiday, Holiday, holidays, xmas, christmas, x-mas — each sitting on a handful of assets, none of them finding the others. Nobody built it on purpose; it’s just what uncontrolled tagging becomes. The fix isn’t nagging people to be more disciplined — it’s a controlled list, so the discipline is built into the tool instead of the human.
Why every tool uses these words differently
Here’s the part nobody warns you about: the confusion isn’t you, it’s the tools. The same three words mean different things in every app, and once you see it, the whole topic gets easier.
- ACDSee has both Categories and Keywords — and, per its own documentation, both are hierarchical trees you can nest and multi-assign. The popular “categories are broad, keywords are granular” rule is user convention, not something ACDSee’s manual actually says.
- Adobe Lightroom Classic has no “categories” at all. It uses keywords, and its own dialog is literally titled “Create Keyword Tag” — so in Lightroom, keyword and tag are the same thing. What looks like categories — Collections — is a separate, catalogue-only construct that isn’t written into the file’s metadata at all.
- Adobe Bridge calls its nested parent keywords “hierarchical categories,” and has a completely unrelated “Categories” field for legacy Associated Press news codes. Two different “categories” in one app.
- IPTC — the metadata standard your files actually carry — defines Keywords as a flat, free-text list, explicitly not a hierarchy; controlled codes are meant to go in a separate “Subject Code” field.
There’s even a formal trap: in NISO’s standard, a “category” is defined as a grouping that specifically does not form a strict hierarchy — the exact opposite of how ACDSee uses the word. So when two people argue about “categories versus keywords,” they’re often describing different software. The practical takeaway: don’t adopt any one tool’s labels as gospel. Decide what you mean by where-it-lives, what-it’s-about, and free-label, then map those onto whatever your tool happens to call them.
Every team I onboard wastes the first meeting arguing whether something is a “category” or a “keyword,” because their last tool used the words differently. I’ve stopped having that argument. I ask two questions instead: does an asset live in only one of these (that’s a category) or many (that’s a keyword)? And is the list of allowed values fixed (controlled) or anything-goes (a free tag)? Answer those two and the labels sort themselves out, whatever your software happens to call them.
A practical model that uses all three
You don’t have to choose. The libraries that stay findable for years combine the three, each doing the job it’s built for:
- Categories for the one structural home — a shallow tree everyone can browse, three or four levels deep, so every asset has one obvious place.
- Controlled keywords for everything cross-cutting — product, campaign, subject, rights status, colour space — drawn from a governed vocabulary so search actually works. This is where most of the value lives.
- Free tags, sparingly, for the ephemeral — personal and temporary notes, reviewed periodically so the useful ones graduate to keywords and the rest are cleared.
Two rules keep it from turning into chaos. Keep one source of truth: the metadata should live with the asset (written into the file as IPTC/XMP where possible) rather than in three tools that disagree. And write the vocabulary down and give it an owner, so it’s governed rather than improvised. If you want the step-by-step on building that structure, our DAM taxonomy best practices guide is the how-to companion to this one, and a tool that reads and writes your keywords back into the files — see our metadata fidelity ranking — is what keeps the whole model portable.
Sources & references
- ANSI/NISO Z39.19 — Guidelines for Controlled Vocabularies — the standard definitions of controlled vocabulary, keyword, taxonomy and category cited here, accessed July 2026.
- IPTC Photo Metadata User Guide — the definition of the free-text Keywords field and separate Subject Code, accessed July 2026.
- Adobe — Lightroom Classic keywords — hierarchical keywords, the “Keyword Tag” naming, and Collections, accessed July 2026.
- ACDSee — Catalog pane — Categories and Keywords as parallel hierarchical trees, accessed July 2026.
- PhotoLib methodology — how we research and test DAM tools. See our methodology.
Keep reading
FAQ
What is the difference between categories, keywords and tags?
A category is where an asset lives - its place in a hierarchy, usually one main home per asset. A keyword is what the asset is about, ideally chosen from a controlled list so everyone uses the same term, with many keywords per asset. A tag is a free-form label anyone can invent on the spot. In short: categories are for structure and browsing, controlled keywords are for consistent search, and tags are for quick, temporary, personal notes.
Are keywords and tags the same thing?
Loosely, people use the words interchangeably - and some tools do too; Adobe Lightroom's own dialog is titled 'Create Keyword Tag'. The useful distinction is control, not the label: a keyword drawn from a controlled vocabulary keeps your search consistent, while a free tag that anyone can invent leads to 'beach', 'Beach' and 'seaside' all describing the same thing. Treat controlled keywords as your durable search layer and free tags as throwaway notes.
Should I use categories or keywords to organize my photos?
Both - they solve different problems and aren't an either/or choice. Use categories for the single structural home an asset lives in, so people can browse the library and understand it. Use keywords for everything that cuts across that structure - product, campaign, subject, rights - because an asset can carry many keywords but only sits in one category. A library that relies on only one of the two is harder to search than one that uses each for its strength.
Why do ACDSee, Lightroom and my DAM use these words differently?
Because there is no agreed standard for the labels, only for the underlying metadata. ACDSee has both Categories and Keywords, and both are hierarchical trees. Lightroom has no categories and treats keyword and tag as the same thing, while its Collections are a separate catalogue-only construct. Adobe Bridge uses 'categories' for two unrelated things. The IPTC standard your files carry defines Keywords as a flat free-text list. The fix is to decide what you mean by where-it-lives, what-it's-about and free-label, then map those onto whatever your tool calls them.
What is tag sprawl and how do I avoid it?
Tag sprawl is what happens when uncontrolled tags multiply into near-duplicates - 'car', 'cars', 'automobile', 'auto' all describing the same thing - until search stops working because no single term finds everything. You avoid it with a controlled vocabulary: an approved list with one preferred term per concept and synonyms mapped to it. Keep free tags for genuinely temporary notes, and promote any tag that keeps getting reused into a proper controlled keyword.