← All articles

Why your Getty keywords show up red (and what to do about it)

September 1, 2026

You finish keywording a clip, paste your list into Getty's contributor portal, and half of it turns red. The uploader tells you "keywords in red are not found," and moves on.

Nothing is broken, and your clip isn't rejected. But those red keywords are doing nothing for you, and understanding why changes how you keyword everything afterwards.

If you would rather not fight the vocabulary by hand on every clip, that mapping is exactly what MetaMule automates. See how it keywords for Getty and Pond5.

Getty and Pond5 use a controlled vocabulary

Most contributors assume keywords are free text, and that typing more words means more chances to be found. On Shutterstock and Adobe Stock, that's broadly true.

Getty and Pond5 work differently. They match every keyword you submit against a controlled vocabulary: a fixed, curated list of terms their search actually indexes. If your keyword isn't in that list, it renders red, and the clip simply isn't findable under it.

The vocabulary isn't arbitrary. It's curated from real buyer search behaviour, the terms people actually type when they're looking to license footage. That has a useful implication we'll come back to.

Red is about matching, not quality

The most common misreading is that red means "bad keyword." It doesn't. Red means unmatched.

These are all perfectly good descriptions that come up red:

What you wrote What the vocabulary has
wild rabbit Rabbit - Animal
wing spreading Spread Wings
white water Rapids - River
reed Reed - Grass Family
ground Land

Look at that list and the pattern appears. The vocabulary prefers:

Qualifiers are where most red comes from

The single biggest source of red keywords is adjective-plus-noun combinations that feel descriptive but aren't concepts: small crab, gray bird, muddy water, dead grass.

A controlled vocabulary can't contain every adjective-noun pair anyone might think of, so it contains the noun. Strip the qualifier and you usually land on a real term: Crab, Bird, Grass. You lose nothing: nobody searches for "small crab" when they need a crab.

The exception is when the qualifier is part of a species or proper name: Blue Jay, Polar Bear, Great Blue Heron. Those are single concepts, not a colour plus an animal.

Some things genuinely aren't in the vocabulary

Sometimes there's no canonical equivalent and the honest answer is to drop the keyword.

A useful example: debris. On a mudflat covered in fallen leaves and twigs, "debris" is an accurate description. But the vocabulary's candidates for it are Ruined, Garbage, Wreck, and Rubble, all human wreckage. Forcing your natural scene into one of those is worse than leaving the keyword off: you've made the clip findable under a term that misrepresents it, which wastes a buyer's click and can flag the asset as mislabelled.

A wrong keyword is worse than a missing one.

How to find the right term

Getty's contributor portal will tell you, if you ask it. Click a red keyword and it offers suggestions from the vocabulary. That's the authoritative answer, straight from the index you're trying to enter.

Two other reliable sources:

  1. Your own published clips. Once an asset is live, its public page lists the keywords Getty accepted and normalized. That's ground truth.
  2. Competitors' published clips in your subject area. If a well-performing clip of a similar subject uses a term, it's in the vocabulary.

Why this matters on Adobe and Shutterstock too

Here's the part most contributors miss. Shutterstock and Adobe accept free text. They'll never show you a red flag. That sounds more forgiving, and in one sense it is.

But it also means you get no feedback. A wasted keyword on Adobe looks exactly like a good one. And since the controlled vocabulary was built from what buyers actually search, those canonical terms are usually the stronger keywords everywhere.

Writing to the vocabulary isn't a Getty-specific chore. It's a discipline that makes your metadata better on all four marketplaces. You just only get graded on two of them. That discipline is built into MetaMule, which normalizes to the canonical terms and exports each marketplace's CSV in its own format.

The short version


MetaMule writes to the vocabulary so you don't see red

It normalizes keywords to the terms Getty and Pond5 actually index, from a mapping built by reading real published clips and the portals' own suggestion lists, then exports each marketplace's CSV in its own format.

See how MetaMule works