Half the catalogue stops at exactly ten tags

Measured 19 September 2026 across 6,519 active listings, 48.20% carry exactly ten tags. No other count comes close. That is not a preference shared by half the shelf — it is the shape of the box they typed into.

Where the tag counts sit

• 0 tags — 759 listings (11.64%)
• 1 tag — 496 listings
• 5 tags — 407 listings
• 7 tags — 463 listings
• 9 tags — 223 listings
• 10 tags — 3,142 listings (48.20%)
• 11 tags — 11 listings
• 12 tags — 78 listings
• 16 tags — 1 listing

Median 9, mean 7.00. A distribution that rises gently and then puts nearly half its mass on one value is describing a limit rather than a behaviour.

The listings above ten say where the limit is

90 listings carry more than ten tags. 89 of them — 98.89% — were created through the API.

The dashboard form caps tagging at ten. The API does not. So the pile-up at exactly ten is the ceiling of an input box, and the few listings above it are the ones that never passed through that box. Whatever half the catalogue wanted to say about itself, ten is where it was asked to stop.

That is worth knowing in both directions. If you tag through the dashboard and feel you are being cut off, you are — and the tenth tag is not a natural stopping point, it is the last slot. If an eleventh tag genuinely changes how your work is found, the API is where it fits.

One listing in nine has no tags at all

11.64% of live listings carry zero tags. That is a larger share than any single count between one and nine, which makes "none" the second most common answer on the whole distribution.

A listing with no tags is not invisible — its title, description and category still do work. But it has opted out of every path that begins with a word somebody typed rather than a category they clicked. Adding three accurate tags takes about a minute and is the cheapest available change to a listing.

What the popular tags actually measure

The most common tags across live listings:

• blender — 539 listings (8.27%)
• textures — 537 · pbr texture — 535 · 3d material — 535 · seamless — 535
• tileable — 464 · nature — 277 · architecture — 235
• building — 231 · brick — 229

Read that list carefully before drawing anything from it. Those tags come overwhelmingly from one well-stocked shelf of PBR material listings, where the same ten words are applied consistently across hundreds of entries. "Popular tag" here means "one thoroughly tagged category", not a trend across the catalogue.

The caveat is the useful part. Every tag-frequency table from every marketplace has this problem, and most published ones do not mention it. If you pick tags by copying what is common, you are copying one category's vocabulary — which helps if you are in that category and does nothing whatsoever if you are not.

What to do with ten

Spend the ten on distinctions, not synonyms. textures, texture and texturing are one tag in three costumes. Use the slots for the software, the format, the use case, the style — four different questions rather than one question four times.

Start with words that are not already in your title. The title is indexed. A tag repeating it buys nothing. A tag naming the tool that opens the file buys a whole search that would otherwise miss you.

If you have none, add three. Not ten. Three accurate tags beat ten guessed ones, and against one listing in nine currently carrying none, three is the entire distance between opted out and findable.