The Steam Workshop's Dead-End Tags (2026): 240 Labels Hold 1.86 Million Mods — and Just 26,637 of Them Reach 1,000 Subscribers

The Steam Workshop's discovery layer is not the homepage, and it is not the leaderboard. It is the tag row — Map, Vehicle, Roleplay, Build 42, savegame — the handful of labels a creator attaches to an item and every player uses to filter a game's catalogue down to the kind of thing they want. ModSignal tracks 110,890 of those tags across 392 games. Which means the tags themselves can now be measured the same way mods are: as markets, with a size and a conversion rate.

Read that way, the Workshop has a set of shelves that almost nobody buys from. Take every tag carrying at least 1,500 mods, and count how many of those mods ever reach 1,000 subscribers. Two hundred forty of them land under 3%. Together they hold 1,856,272 mod-slots and produced 26,637 hits — roughly one breakout per 70 mods, in the tags where creators put the most work.

The shelf that never sells

The plainest examples are in small games, where a single tag can be a third of the catalogue.

Nothing in this table is a rounding error. Modded Buses is a shelf of 23,052 items with 18 that ever crossed a thousand installs. Diorama and autoplay are perfect zeros. And five of the eight are inside Poly Bridge 3 — a game where the same creators, working the same tools, also produced hydraulic-controller, the title's highest-converting label.

That last detail is the whole point, and it is worth slowing down for.

The game is not the answer — the tag is

The easy dismissal is that these are bad games. Poly Bridge 3's overall catalogue reaches 1,000 subscribers just 0.23% of the time; BattleBlock Theater sits at 0.34%, From the Depths at 1.33%. Surely the dead tags are just the dead games showing through.

They are not. Sort the tags by how far each one deviates from its own game's baseline, and the two stop moving together.

Inside Transport Fever 2 — a healthy ecosystem where 41.63% of all mods reach a thousand — the savegame shelf holds 5,296 mods and reaches 1.25%. Inside RimWorld, a 29.33% game, the Scenario tag manages 1.48%. The game is fine; the shelf is not.

The reverse shows it just as cleanly. Prison Architect is a weak ecosystem overall — 2.34% reach. But the tag literally called content runs 50.56%, and plain mod runs 45.66%. In Besiege, where the game average is 3.05%, the New Textures shelf hits 45.88% and Skin Packs — with 3,741 mods — reaches 32.85%. In City Bus Manager, a 1.88% game, Solo reaches 47.58%.

Across the 2,273 (game, tag) shelves large enough to measure, 138 sit 3× or more below their game's reach and 166 sit 3× or more above it. Roughly 13% of all shelves break cleanly away from the ecosystem they live in. The tag is carrying information the game does not.

The full market ladder

Zoom out and the shape is not a simple "bigger shelf, worse conversion." It is a plateau with a dip — the middle of the market is the worst place to publish.

The most crowded shelves are not the dead ones. The 5,000-to-20,000 band is — 18.72%, the floor of the ladder — while the giant 20,000-plus shelves rebound to 24.99%. Size alone does not kill a tag; a large, generic, well-understood label (Mod, Roleplay, Vehicle) still converts. What kills a tag is specificity without demand: a narrow shelf big enough to be a real market, crowded enough to bury an item, and shallow enough that nobody browses it.

The distribution is that stark. Bucket the tags of 200-plus mods by conversion:

A thousand tags are effectively dead (under 5%); a hundred are gold. It is an all-or-nothing market, and the label a creator chooses decides which side of it they land on.

Why the shelf goes dead

Not every graveyard tag is a creator's mistake. Several are the games' own file-format vocabulary — savegame in Transport Fever 2, Wallpaper Wall in Rust — shelves that were never meant to convert but to classify. That distinction matters, because it also tells a creator where the mistake lives. If a tag exists to sort inventory, it is not a market; if a tag exists to describe a kind of mod, it is one, and its conversion rate is a fact about that kind of mod you can look up before you ship.

What the numbers cannot tell us is direction. A dead tag may be a signal that only low-effort mods use that label; it may equally be that players simply never open it. The honest reading is the conservative one: the label is associated with a 3× swing in reach that survives holding the game fixed, and the shelf's conversion rate is knowable in advance. That converts a guess into a query.

What it means for a creator

The practical version is unglamorous and hard to argue with. Before shipping into a game, pull the tags that fit the mod and read their conversion. A shelf holding twenty thousand items and returning one breakout per eighty is not a shelf you want your work filed under — and it is often a shelf the same game offers a better neighbour to. Poly Bridge 3 will file a build tool under springs (0.119%) or under hydraulic-controller (0.202%); the catalogue is the same, the game is the same, and the odds are not.

The Workshop has spent a decade teaching creators that the game is the market. At the level of the tag, that stops being true. There are 240 shelves on the platform that have quietly become graveyards — 1.86 million mod-slots, 26,637 installs past the thousand mark — and most of the items in them were filed there on purpose by people who never had a way to check.

Figures are a single read of ModSignal's live Workshop dataset: 6,017,047 active, subscribed Steam Workshop mods across 392 tracked games and 110,890 distinct tag labels. A "hit" is an item with 1,000 or more current subscribers; "reach" is the share of a tag's items that are hits. Shelf-level comparisons hold the game fixed and require at least 500 items per tag and 1,000 per game; the graveyard cut is every tag with 1,500+ items converting under 3%.