Two layers: subject and meaning
A good keyword set always has two parts.
The literal layer is what's physically in the frame: subject, place, action, colour, angle. The meaning layer is what the shot stands for and what it would be bought for: an emotion, an idea, a use case.
Literal words help people who already know what they're looking for. Meaning words bring the buyer who needs a picture to go with a point — and that's where most sales come from.
A worked example
Take this shot: a woman working on a laptop in a café by the window, morning light, a cup of coffee on the table.
Literal layer: woman, laptop, café, work, window, morning light, coffee cup, casual clothes, one person, side view.
Meaning layer: remote work, freelancing, productivity, self-employment, digital nomad, quiet morning, work-life balance.
Longer phrases: "freelancer working from a café", "remote work early in the morning", "woman with a laptop by a window".
Final order — strongest first: woman with laptop, café, remote work, freelancing, morning light, window, coffee cup, casual clothes, productivity, digital nomad, quiet morning.
The main subject and the main concept go in the first positions; everything else follows in descending order of importance.
Work both sides
Pond5 suggests tagging the opposite associations as well as the obvious ones. A shot of fried bacon is "tasty, breakfast, appetising" and also "unhealthy eating, cholesterol, bad habit". One file gets licensed for a recipe piece and for a heart-health piece.
It opens up searches you'd never reach by only asking "what is this a shot of".
Building longer phrases
Broad words like "business" or "nature" compete with millions of files. A long, precise phrase competes with almost none.
A simple pattern: who or what + doing what + where or how.
"An elderly couple walking through an autumn park", "a team discussing a project at a whiteboard", "a drone flying over a mountain lake at dawn".
Two or three phrases like that, plus a handful of single words for the broad category — that's the working balance.
How many keywords
An image usually needs 15–25; video needs more, because you also have to describe the camera movement, the motion in the frame and the shot type.
The cue isn't the platform's limit — it's the point where your accurate words run out. Anything after that is padding: it dilutes relevance, and on some platforms it counts as spam.
Check yourself
A quick checklist before you upload:
- Would a buyer type this into search?
- Is it visible in the frame — directly, or as an obvious meaning?
- Is the strongest term first?
- Any words that repeat one meaning in different forms?
How Meta VisBreeze handles it
The generator builds both layers for you. Your rules are applied automatically: required words, stop-lists, trademark removal. The keyword count, format and order are fitted to the platform you pick.
The list stays fully yours to change — edit it across the batch or one file at a time.
FAQ
Where do I start if I'm stuck?
Describe the shot out loud in plain words, as if over the phone: who, doing what, where, what mood. That's your first draft of the keywords — then trim the excess and add the use cases.
Do they have to be in English?
It depends on the platform: many search an English index, some accept local languages. Latin species names, place names and established terms stay as they are.
What goes in the keywords for an abstract or minimalist piece?
Lean on the meaning layer and on use: "presentation background", "cover art", "space for text", plus mood and colour palette.
Does keywording change for AI images?
No. Keywords describe what's visible in the frame — how the file was made doesn't affect that. Platform rules are a separate matter: some don't accept AI content, or require it to be labelled.