Most Steam Workshop analysis treats the platform as one market. It is two, and the difference is a single field Steam's API returns for every listing: file_type.
Type 0 is Community — an ordinary mod, open to anyone. Type 1 is Microtransaction — content submitted and voted on for the purpose of being sold in-game. The curated lane exists in just 12 game ecosystems; the community lane exists in 396.
The obvious question is which lane is worth being in. The answer is "community," in almost every ecosystem on Steam — and then there is Rust, where the answer reverses completely and takes the intuitive story with it.
The two lanes, sized
| Tier | Items | Share of catalogue | Items with ≥1 subscriber | Reaching 1,000 subscribers |
|---|---|---|---|---|
| Community (type 0) | 6,781,829 | 92.22% | 5,993,835 | 14.16% |
| Microtransaction (type 1) | 369,523 | 5.02% | 4,697 | 1.12% |
| Unclassified | 202,623 | 2.76% | 0 | 0.00% |
The curated lane's central fact is not its subscriber share — that story has been told, and it holds up. It is the column that has been sitting in the data unread: of 369,523 curated items, only 4,697 have ever registered a single subscriber. That is 1.27%.
Community mods register at least one subscriber 1,276 times as often.
So the pooled headline writes itself: the curated lane is a ghost town. It reaches 5% of the catalogue's items and 0.46% of its subscriptions.
But pooled numbers across 12 ecosystems are almost always hiding something, and this one is hiding an inversion that runs the other way.
The condition that flips the picture
Stop counting all items and count only the ones that found any audience. The curated lane's problem is not conversion — it is entry.
| Tier | Items with ≥1 subscriber | Of those, reaching 1,000 | Rate |
|---|---|---|---|
| Microtransaction | 4,697 | 4,130 | 87.93% |
| Community | 5,993,835 | 848,654 | 14.16% |
Among items that got off zero at all, curated items clear 1,000 subscribers 6.2× more often than community mods.
That is not a fluke and it is not small. It is what you would expect if the curated lane were a filter: a tiny number of items pass a gate, and the ones that pass have already been voted on and selected. The community lane has no gate at all, which is why 30% of it carries no tags and a tenth of it is parked at exactly one subscriber.
The honest reading is that the curated lane fails at distribution — almost nothing reaches players — while succeeding at selection. Those are different failure modes and the pooled table cannot tell them apart.
Then you split it by game and the story breaks
Here is the same comparison run inside each of the 12 games that operate a curated lane. The community column is that game's own mods; the curated column is that game's own items.
| Game | Community mods | Reach 1,000 | Curated items | Reach 1,000 |
|---|---|---|---|---|
| Wallpaper Engine | 106,060 | 93.718% | 2 | 0.000% |
| Company of Heroes 2 | 16,005 | 41.912% | 60 | 0.000% |
| Counter-Strike 2 | 31,250 | 26.560% | 115,320 | 0.000% |
| Primal Carnage: Extinction | 27 | 25.926% | 6,030 | 0.000% |
| Dota 2 | 7,825 | 25.304% | 44,861 | 0.000% |
| Killing Floor 2 | 3,519 | 23.245% | 1,174 | 0.000% |
| A Hat in Time | 5,936 | 22.439% | 28 | 0.000% |
| Team Fortress 2 | 12,719 | 9.710% | 37,329 | 0.000% |
| Unturned | 113,801 | 3.587% | 3,901 | 0.000% |
| SpeedRunners | 38,564 | 1.955% | 473 | 0.000% |
| Rust | 19,997 | 0.135% | 159,165 | 2.595% |
| PAYDAY 2 | 0 | — | 1,182 | 0.000% |
Read the right-hand column. In 11 of the 12 games, curated items reach 1,000 subscribers at exactly 0.000%. Not "rarely." Zero, to three decimal places, across 208,600 items in those 11 ecosystems.
Counter-Strike 2 publishes 115,320 curated items and not one has reached a thousand subscribers. Dota 2 publishes 44,861 and reaches zero. Team Fortress 2 publishes 37,329 and reaches zero. These are the three largest curated catalogues on Steam outside Rust, and collectively they are 197,510 items with an identical result.
And now Rust
Rust is the twelfth game, and it is the only row on the table where the two columns swap places.
Rust's curated items reach 1,000 subscribers at 2.595%. Rust's community mods reach 1,000 at 0.135%.
The curated lane is 19× healthier than Rust's own mod lane. That is the finding, and it inverts the pooled story so completely that both numbers belong on the same line.
Three more cuts make it starker.
Rust's mod lane has no top end at all. Those 19,997 community mods produce exactly 27 items above 1,000 subscribers and zero above 10,000. The single most-subscribed Rust mod on the Workshop has 5,191 subscribers. Rust's most-subscribed curated item has 82,466 — sixteen times larger. The 90th percentile of Rust mods sits at 3 subscribers.
Rust's curated items reach 1,000 at 88.06% among those that found any subscriber — the figure that makes the tier look like a filter rather than a lottery. Every other game with a subbed curated population sits at 0.00%, on populations so small that the rate is meaningless: Team Fortress 2 has 4, Dota 2 has 1, PAYDAY 2 has 2.
Rust's curated lane carries 98.2% of Rust's entire subscriber base. 199,997 mods and 159,165 curated items, and the curated side holds nearly all of the installs.
Compare the top Rust mods by name and the mechanism becomes obvious. The leaders read +5 SKINDROP RUSTILA, Golden Unlock, gift, 3484167544, Beans Unlock, hazmatus. Rust's community Workshop lane is dominated by unlock-abuse and low-effort gift uploads, not by content anyone curates. In Rust, the mod lane is the noise — and the curated lane is where the actual content economy lives, because that is where skins go before they rotate into the in-game store.
That is the same conclusion the earlier analysis of this tier reached from the other direction, and it survives being tested per-game rather than pooled: Rust built a working loop and everyone else built a display case.
The spread across ecosystems
The table above also exposes a second, quieter finding that has nothing to do with curated tiers. Look at the left column — the reach rate of each game's ordinary community mods:
Rust 0.135% ... and Wallpaper Engine 93.718%.
That is a 694× spread between the weakest and strongest community mod lanes among these ecosystems. Company of Heroes 2 sits at 41.9%, Counter-Strike 2 and Dota 2 at a quarter of their catalogues each, Team Fortress 2 at 9.7%, and then a long decline to SpeedRunners at 1.955% and Rust at 0.135%.
A "does this mod reach 1,000 subscribers" number means something completely different depending on which ecosystem you are standing in. In Wallpaper Engine, essentially every mod that exists reaches 1,000 — the Workshop page is the distribution channel. In Rust, essentially no mod does, because the Workshop page is not where Rust's content is distributed at all — the curated lane is.
Anyone benchmarking a mod's performance against "the Steam Workshop average" is comparing across a 694× range and calling it a baseline.
What the two lanes do over time
Both lanes are getting harder over time, and the curated lane harder still.
| Year | Curated items | Reach 1,000 | Community mods | Reach 1,000 |
|---|---|---|---|---|
| 2019 | 10,742 | 3.30% | 295,355 | 18.29% |
| 2021 | 27,516 | 1.68% | 528,878 | 16.03% |
| 2023 | 41,152 | 1.18% | 764,314 | 12.17% |
| 2025 | 62,956 | 0.90% | 1,121,692 | 9.68% |
| 2026 | 42,628 | 0.95% | 975,632 | 6.74% |
Curated-item publishing volume has grown 6× since 2019 while its reach rate has fallen to a third of what it was. The community lane shows the same decay at a gentler slope — 18.29% to 6.74%.
Note where the curated lane's volume actually goes. Of 369,523 items, 115,320 are Counter-Strike 2 and 44,861 are Dota 2 — 43% of the entire tier lives in two ecosystems where its reach rate is zero. The tier is not growing because it is working. It is growing because two games keep minting items into a lane nobody installs from.
The creators who tried both
One more control, because "curated items work in Rust" could be a statement about Rust's creators rather than its tier. It is not.
828 creators have published both a community mod and a curated item in the same game. Their community mods reach 1,000 subscribers at 14.92%. Their curated items reach 1,000 at 0.00% — 0 of 15,619.
Same people, same game, same period. The lane is the variable.
Sorting all creators by which lanes they use shows the same thing from another angle:
| Creator profile | Creators | Their curated items reaching 1,000 |
|---|---|---|
| Items only | 28,137 | 1.24% |
| Cross-tier (both lanes) | 3,524 | 0.29% |
The 28,137 creators who work only in the curated lane do four times better within it than the 3,524 who also make mods. Specialists beat generalists in a lane that barely functions — which is what you would expect if reaching players there requires treating it as a job rather than a sideline.
What this means
For creators. The lane choice is not a style choice, it is an ecosystem fact you can look up. In 11 of 12 curated ecosystems the correct answer is "publish a community mod," and the margin is not close — it is 197,510 curated items with a 0.000% reach rate against community lanes running from 2% to 94%. In Rust the correct answer is the opposite, and the reason is structural: Rust routes its skins through the curated lane into an in-game store, so the lane is the content economy. Ask which lane your target game actually distributes through before you pick one, because the platform will not tell you.
For anyone benchmarking. The pooled statistic — "curated items reach 0.45% of subscriptions" — is true and useless on its own, because the same dataset contains a lane that reaches 98.2% of its game's installs and eleven lanes that reach zero. Aggregate rates across a 694× spread are not a baseline, they are an average of incomparable things.
The general lesson. The natural summary of the curated tier is "the dead lane." That summary is correct in 11 of 12 ecosystems, and it has been published as the conclusion. But it was produced by pooling, and pooling is what made it look like a property of the tier instead of a property of the eleven games that happen not to route content through it. The tier's own behaviour, conditioned on the games that use it properly, looks nothing like its average. Rust is not the exception that proves the rule — Rust is the case that shows the rule was never about the tier.
Where the numbers come from
A single snapshot of ModSignal's live Steam Workshop database on 28 September 2026: 7,353,981 active items across 396 game ecosystems, of which 369,523 carry file_type = 1 (Microtransaction) across 12 games, and 6,781,829 carry file_type = 0 (Community). A further 202,623 items carry no file_type and are excluded — they have no creator attributed and no subscription at all.
"Reaching 1,000 subscribers" means subscriptions_current >= 1000, an install counter Steam reports per item. The per-game table is restricted to the 12 ecosystems that operate a curated lane; PAYDAY 2 publishes no community mods in the window, so its community column is empty rather than zero. Rates for very small populations should be read as directional: four ecosystems have fewer than 1,200 curated items, and Primal Carnage: Extinction's 27 community mods cannot carry a confident percentage.
One caveat follows this dataset everywhere and matters here. ModSignal samples the catalogue on a tiered cadence rather than snapshotting every item continuously, and the pipeline deliberately keeps low-traffic items without stripping their counters — so the "reach 1,000" rates reflect the tracked population, not a claim about every item Steam has ever hosted.
The invisible second tier analysis covers the pooled version of this dataset; this piece re-cuts it per ecosystem. For the lane that carries the platform, see the one-subscriber floor and the churn economy; for what predicts whether a given mod gets traction in the first place, the debut signal and the portfolio effect are the relevant controls. Game pages with full catalogues: Rust, Counter-Strike 2, Dota 2, Team Fortress 2, Wallpaper Engine, Unturned, Company of Heroes 2, Killing Floor 2, SpeedRunners, A Hat in Time, PAYDAY 2 and Primal Carnage: Extinction.