Shutterstock Metadata Generator

The seven Shutterstock columns

The Shutterstock export carries a wider row than Adobe's. Every submission is written with these seven columns:

The file downloads as Shutterstock_CSV_<DD-MM-YYYY>.csv and can hold up to 2,000 rows. Column headers and a sample row are listed on the templates and column reference, which is worth checking if you import the CSV somewhere else first.

Categories are classified from the image

Category picking is the part of Shutterstock metadata that slows people down, so Metadata Studio handles it directly. For each image, a separate vision call looks at the picture and classifies it against the 26 official category names. That classification is a distinct request: it never changes the title, description or keyword generation, and if it fails the rest of the metadata is unaffected.

The suggested category is a starting point, not a verdict. Every value stays editable in the table, so you can change it or clear it before export. You can also add a second category when an image genuinely sits in two groups.

The 26 canonical categories

The classifier chooses only from this fixed list, which is why the names in your CSV always match what Shutterstock accepts:

• Abstract
• Animals/Wildlife
• Arts
• Backgrounds/Textures
• Beauty/Fashion
• Buildings/Landmarks
• Business/Finance
• Celebrities
• Education
• Food and Drink
• Healthcare/Medical
• Holidays
• Industrial
• Interiors
• Miscellaneous
• Nature
• Objects
• Parks/Outdoor
• People
• Religion
• Science
• Signs/Symbols
• Sports/Recreation
• Technology
• Transportation
• Vintage

At least one valid category is required per row. If a row has none, the export returns an error and names the offending files instead of producing a CSV the marketplace would reject. Fix the named rows and export again.

Editorial, Mature Content and Illustration

The three flag columns take a simple Yes or No. Editorial tells Shutterstock that an asset is editorial rather than commercial, and the tool warns you when it is set: editorial assets need a capture date and location in the description, so the caption should say when and where the image was made. Mature Content and Illustration flag the other two special cases.

There is a second, separate warning when a row has fewer than 20 keywords. It is non-blocking — the export still runs — but it is a useful nudge, because a thin keyword list is one of the easiest ways to lose relevant searches.

Metadata limits and export

Descriptions are capped at 150 characters and keywords at 49, with 20 to 40 the recommended range. The category classifier and the metadata model are both driven by the same uploaded image, but they run as independent calls, so a slow classification never blocks the text fields.

When the table is ready, switch to the Shutterstock tab and export. The relay writes the file in your browser under the Shutterstock_CSV name, and your images and API key never leave your machine. To understand the underlying model, see the AI generator overview and the six-step workflow.

A tidy Shutterstock pass

1. Upload the batch and let each image generate its description, keywords and category.
2. Check the suggested category against what you know about the shoot, and correct it where needed.
3. Confirm every row has at least one category so the export does not stop.
4. Set the Editorial, Mature Content and Illustration flags and add a capture date and location for editorial frames.
5. Export the CSV and submit it.