Keywording is the practice of adding descriptive tags to an asset — the subject, the setting, the people, the use case — so it can be found later by searching for those words, instead of relying on the person looking to remember which folder it was filed in.
In plain English
A folder structure only works if you remember, months later, exactly which folder you filed something in. Keywording solves the more common real problem: you remember what's in the photo — "the launch event," "the CEO," "the blue product shot" — but not where it lives. Adding those descriptive words to an asset as searchable keywords means anyone can find it later by typing what they remember, regardless of the folder it's actually stored in.
Done manually, keywording is one of the most time-consuming parts of running a library — which is exactly why auto-tagging exists: AI that suggests keywords automatically so a person doesn't have to type them one asset at a time. But even with auto-tagging doing most of the volume, someone still needs to keyword the business-specific details a generic AI model can't guess — a campaign name, a client name, an internal project code.
Left completely unstructured, keywording drifts: one person types "dog," another types "puppy," a third types "canine," and now the same real-world subject is split across three unrelated search terms. That's the problem a controlled vocabulary solves — an approved list everyone keywords from, so synonyms collapse into one consistent term instead of multiplying.
Why it matters in a DAM
Search is only as good as the keywords behind it. A beautifully organized folder tree still fails the moment someone searches instead of browses, if the assets underneath were never keyworded. Consistent keywording is what lets a library scale past the point where any one person remembers where everything is filed — which, in practice, is a fairly low bar most libraries cross within their first year.
Buyer’s test: ask how the tool handles keywording at scale — batch-applying the same keywords to dozens of selected assets at once, not just one at a time (a batch operation, applied to keywording). A tool that only supports single-asset keywording will make a large backlog genuinely impractical to catch up on by hand.
Related terms
See it in action
Our photo library organization guide covers building a keywording workflow from scratch, including a worked category structure. For a tool with fast batch-keywording tools built specifically for large photo libraries, see our Daminion.
FAQ
What is keywording in digital asset management?
Keywording is the practice of adding descriptive tags to an asset - the subject, the setting, the people, the use case - so it can be found later by searching for those words, instead of relying on someone remembering which folder it was filed in. It solves the common real problem: you remember what is in the photo, not where it lives. Adding those words as searchable keywords means anyone can find the asset by typing what they remember, regardless of where it is stored.
How is keywording different from auto-tagging?
Keywording is the general practice of adding descriptive tags, whether a person types them or a machine suggests them. Auto-tagging refers specifically to AI proposing those keywords automatically. Auto-tagging reduces manual keywording rather than eliminating it: a generic model can recognize a subject in the frame, but it cannot guess the business-specific details that make an asset findable inside your organization - a campaign name, a client name, an internal project code. Someone still supplies those.
Why do keywords drift, and how do you stop it?
Left unstructured, keywording drifts because people describe the same thing differently. One person types 'dog', another 'puppy', a third 'canine', and the same real subject is now split across three unrelated search terms that no one will think to combine. A controlled vocabulary is the fix: an approved list everyone keywords from, so synonyms collapse into one consistent term instead of multiplying. Without it, search quality quietly decays as the library and the number of people adding to it grow.
Where do keywords actually get stored?
It depends on the tool, and the difference matters. Keywords can be written into the file's own embedded metadata, in which case they travel with the asset when it is copied, downloaded or imported into a different system. Or they can live only in the DAM's own database, tied to that tool. Both are common. If you ever expect to migrate or export a library, ask which one your tool does, because database-only keywords are exactly what gets silently lost in a move.
How do I keyword a large existing backlog?
Look for batch operations first. The question worth asking is whether the tool can apply the same keywords to dozens or hundreds of selected assets at once, not just one asset at a time. A tool that only supports single-asset keywording makes a large backlog genuinely impractical to catch up on by hand, no matter how good its search is afterward. Auto-tagging can cover volume on generic subjects, but batch editing is what lets a person add the specifics efficiently.