Keywording is the unglamorous step that decides whether a strong stock photo ever gets seen. On every major agency, buyers find images through search, and search runs on your title, description, and keywords. A technically perfect frame with vague or careless keywords simply never surfaces. This guide lays out a repeatable stock photo keywording workflow: how agencies actually use your keywords, a batch process you can run on every upload, the mistakes that trigger rejections, and where AI-assisted search genuinely saves time.
How stock agencies use your keywords
Most agencies rank a photo for a search by comparing the query against three fields you control: the title, the description, and the keyword list. Two details matter more than photographers expect.
First, order is weighted. Many platforms give more importance to your first keywords, so the most relevant terms should come first, not whatever you happened to type last. Second, relevance beats volume. Padding a list with loosely related words ("beauty", "lifestyle", "success") to fill every slot dilutes your relevance score and can get a submission flagged as keyword spam.
Limits and conventions vary by agency and change over time, so always confirm the current guidelines for each platform you submit to. As a general picture:
| Field | What it does | Practical guidance |
|---|---|---|
| Title / caption | Primary human-readable summary; often weighted heavily in search | One clear sentence describing subject, action, and setting. Avoid camera jargon. |
| Description | Extra context for buyers and search | Expand on the scene and possible use cases without stuffing keywords. |
| Keywords | The structured tags search matches against | Order by importance; lead with literal subject terms, then concepts. Drop anything not actually in or about the image. |
A repeatable keywording workflow
The fastest way to keyword consistently is to run the same short process on every batch instead of improvising per image.
- Select before you keyword. Decide which frames you are actually submitting first. Keywording your entire shoot wastes the most time of anything in this process.
- Describe literally, then conceptually. Start with what is unambiguously in the frame: subject, action, location, count, colors. Only then add conceptual terms like mood, season, or theme that a buyer might search.
- Order by importance. Put the strongest, most specific terms first. "Golden retriever" before "dog", "dog" before "animal".
- Write the title and description. Keep the title to one plain sentence. Note model or property release status in your records where relevant.
- Prune for relevance. Re-read the list and cut anything that is not literally in or directly about the image. This single pass prevents most keyword-spam rejections.
Common mistakes that waste time or cause rejections
- Over-tagging. Filling every keyword slot with weakly related words lowers relevance instead of raising reach.
- Copy-pasting one keyword set across a whole shoot. Near-duplicate frames still differ; identical tags hurt the ones that don't match.
- Conceptual terms with no visual basis. "Freedom" on a plain product shot reads as spam to review systems.
- Ignoring the order. Burying your best term at position 40 wastes the weighting agencies give early keywords.
Where AI helps with keywording, and where it doesn't
AI keywording tools can draft a long list of tags from an image in seconds, which is a real time saver for the first pass. The catch is the same one human over-taggers fall into: automated tools tend to suggest too many terms, including generic or off-topic ones. The reviewing still has to be yours. Treat AI output as a draft to trim, not a final list to submit.
There is a second, often more useful role for AI in this workflow: not generating tags, but finding the right images to submit in the first place. If you can search your library by what a photo shows, you spend your keywording time only on frames worth uploading.
Find the shots worth submitting
This is where a local-first photo manager earns its place in a stock workflow. Memora runs entirely on your own computer and uses AI to understand the visual content of your photos, so you can locate candidates without having pre-tagged everything by hand.
A few current Memora features map directly onto the keywording problem:
- Semantic search lets you type a plain description — "red umbrella on a rainy street", "team meeting in a bright office" — and surface matching frames instantly, so you can pull submission candidates by concept before you write a single keyword.
- Smart Albums automatically sort imported photos into browsable categories such as Portraits, Landscapes, Architecture, and Food, which makes triaging a large shoot for stock-worthy subjects much faster.
- Automatic captions and keywords are generated locally for every photo, giving you a starting description to refine rather than a blank field.
- Lightroom and Capture One catalog import brings in your existing ratings, labels, and keywords, so prior organizing work is preserved instead of repeated.
Because all of this processing happens on your machine, your unpublished library never gets uploaded to a third-party service just to be searched. You then enter the final, curated title, description, and keywords in your stock agency's portal or upload tool as usual. For a closer look at how machine search compares with manual tagging, see Image Keywording in 2026: Manual Tags vs AI Search.
Pre-submission keywording checklist
- Only the frames you are actually submitting are keyworded.
- The most important, most specific terms come first.
- Every keyword is literally in, or directly about, the image.
- Title is one clear sentence; description adds context, not keyword stuffing.
- No near-duplicate frame shares an identical tag set.
- Release status is recorded where people or private property appear.
Frequently asked questions
How many keywords should a stock photo have?
Enough to cover the subject accurately and no more. Quality and relevance matter more than hitting a maximum. Many strong submissions sit comfortably in the range of roughly fifteen to thirty precise keywords; padding beyond what the image supports tends to hurt rather than help. Check each agency's current limit, since they differ.
Are AI keywording tools worth using?
For a fast first draft, yes — but always trim the output. Automated lists lean toward over-tagging, and irrelevant keywords can get a submission rejected. The bigger AI win is search: finding which images to submit, not just generating their tags.
Do keywords actually affect how my photos rank on stock sites?
Yes. Search relevance is built from your title, description, and keywords, and several platforms weight earlier keywords more heavily. Accurate, well-ordered keywords are a direct lever on discoverability.