The Steam Workshop's own advice column has a favourite topic, and it is the clock. Publish at the right hour. Avoid the evening crush. Catch the quiet window before the Americans wake up.
We tested that advice and it does not hold up. But testing it surfaced something that does, and it is the thing the advice was always gesturing at without ever naming it.
The Workshop's real constraint is not when you publish. It is how much else lands with you.
Using ModSignal's live dataset — 7,324,529 active Workshop items, 12.57 billion subscriptions and 2,058,201 distinct creators — we took every active, publicly-listed mod published between January 2022 and August 2026: 4,591,447 mods. For each one we counted how many other mods entered the same game in the same clock hour. Not the same hour across the whole platform — the same hour inside the ecosystem that mod actually competes in. That count is what a browsing player's front page actually contains.
The crowding curve
| Rivals shipping into the same game-hour | Mods | Share of cohort | Reached 1,000 subs | Reached 10,000 subs | Avg subs | Share of all subscriptions |
|---|---|---|---|---|---|---|
| 1 (alone) | 949,173 | 20.7% | 14.14% | 2.80% | 1,663 | 30.6% |
| 2–3 | 1,170,075 | 25.5% | 11.60% | 2.32% | 1,511 | 34.3% |
| 4–10 | 1,581,404 | 34.4% | 9.51% | 1.48% | 930 | 28.5% |
| 11–50 | 819,132 | 17.8% | 4.24% | 0.50% | 385 | 6.1% |
| 51–200 | 62,445 | 1.4% | 3.64% | 0.64% | 351 | 0.4% |
| 201+ | 9,218 | 0.2% | 6.90% | 1.63% | 605 | 0.1% |
The gradient runs almost monotonically from one end to the other. A mod that arrives alone into its game-hour clears 1,000 subscribers 14.14% of the time. A mod that arrives alongside eleven or more rivals clears it 4.24% of the time. Both figures are large samples — 949,173 mods and 890,795 mods respectively — and the ratio between them is 3.3x.
Follow the subscriptions column instead and the concentration is starker than the rate gap suggests. Mods that shipped into a quiet game-hour (three or fewer rivals) are 46.2% of the cohort and 64.8% of its subscriptions. Mods that shipped into a crowded one (eleven or more) are 19.4% of the cohort and 6.6% of its subscriptions. The 10,000-subscriber column tells the same story, 2.80% falling to 0.50% — a 5.6x spread.
That last row of the table, the 201-plus swarm, breaks the monotonicity, and it is worth naming why rather than hiding it. There are only 192 such game-hours in the entire cohort and they are mostly single mass-upload events — the largest is Castle Crashers on 6 August 2025, where 710 mods arrived in one hour. A handful of those waves are game-update-driven, where thousands of players install everything at once, so the band gets propped up by a few genuine winners. The mass-dump pattern at those volumes is real, but at this tail length the sample is too small and too event-shaped to read.
The hour test, and why it fails
Here is what makes this publishable rather than a restatement. The quiet-hour advice is not wrong on the pooled numbers. Run the hour curve on this cohort and the quiet hours do look better:
| Publish window (UTC) | Mods | Reached 1,000 subs |
|---|---|---|
| 05–11 | 1,022,936 | 11.67% |
| 12–17 | 1,405,606 | 10.33% |
| 18–04 | 2,162,905 | 8.95% |
A 1.30x gap, and it survives every conventional control. Hold the publish year and it holds in all five (2022: 14.81% vs 12.34%; 2023: 12.95% vs 10.61%; 2024: 11.92% vs 8.65%; 2025: 11.77% vs 7.27%; 2026: 8.01% vs 5.33%). Hold the creator's catalog size and it holds in every band. Strip out same-day bursts so a mass-upload cannot masquerade as an hour effect, and it is still 12.21% against 8.99%.
Then run the test that decides it: compare creators to themselves. Take every creator who published into the same game in both a morning window and an evening window, and ask which of their two windows did better:
- 22,625 comparable creator-game pairs with at least three mods on each side.
- The morning window won 48.4% of the decisive pairs.
- Tighten to five mods per side — 10,824 pairs — and it is 48.2%.
The evening slightly wins, at every threshold. The morning advantage is not merely smaller than the pooled 1.30x; it is gone, and faintly reversed. That is the signature of composition: pooled, the clock is a proxy for which creators and which games fall into each window. Held to the same creator and the same game, it stops predicting anything.
The same test, run on crowding, passes
Now run the identical test on the rival count. Same creators, same games, same paired structure — compare each creator-game pair's mods published into a quiet game-hour against their mods published into a crowded one, four to fifty rivals:
- 47,294 comparable creator-game pairs.
- The quiet hours won 55.4% of the pairs, against 44.6% for the crowded ones.
- Tighten to five mods per side — 24,143 pairs — and the quiet rate is ahead 56.3%.
- Widen the crowd definition from 4–50 rivals to 11-plus and the gap grows: 60.4% of 11,318 pairs go the quiet hour's way.
Pool the per-pair hit rates rather than counting wins, and the quiet mods average 12.97% against 12.20% for crowded ones, with mean subscriptions of 1,571 against 1,421.
That is not a large effect. It is a real one, and the contrast is the point: measured inside the same creator and the same game, the hour is a coin flip at 48.4% and the crowd is a 55.4% edge. Run both tests one level up, at the game level across the 46 games with a thousand mods in every cell, and it splits the same way — 51.1% for the hour against 66.7% for crowding. Loosen the cell threshold to 300 mods and 75 games qualify: 48.6% against 64.4%.
The hour is a fingerprint. The crowd is a lever.
Why the two look alike
They look alike because they are entangled, and you can watch the entanglement in the numbers. Split the cohort three ways by publish window and three ways by rival count:
| Publish window | Quiet hours (≤3 rivals) | Mid (4–10) | Crowded (11+) | All mods, pooled |
|---|---|---|---|---|
| 05–11 UTC | 13.13% | 11.68% | 5.79% | 11.67% |
| 12–17 UTC | 13.22% | 10.54% | 4.70% | 10.33% |
| 18–04 UTC | 12.24% | 7.83% | 3.40% | 8.95% |
Read the columns, not the rows. Within the quiet band, the entire hour effect is 13.13% against 12.24% — under one percentage point. Within the crowded band, the hour effect grows to 5.79% against 3.40%. Almost all of the pooled hour gap comes from one fact about the morning window: it contains proportionally fewer crowded hours than the evening does.
Direct-standardise the cohort to remove that. Reweight each window so all three are compared on the identical rival-count mix — 46.2% quiet, 34.4% mid, 19.4% crowded, the cohort average. The hour gap collapses from 1.30x to 11.21% against 9.00%, a 1.25x gap that is now attributable to the residual hour-by-crowd interaction rather than to the clock. Decile-standardise on the rival count instead and it falls slightly further, to 11.17% against 9.00%.
Meanwhile the crowd gap, standardised the same way across publish windows, does not budge. It is not borrowing its signal from the clock.
The mechanism is supply, and it is getting worse
If crowding matters, the interesting question is whether the Workshop is getting more or less of it. It is getting more.
| Publish year | Mods | Mean rivals per publishing hour | Share alone | Share in a swarm (11+) | Reached 1,000 subs |
|---|---|---|---|---|---|
| 2022 | 686,361 | 6.62 | 26.2% | 13.7% | 13.42% |
| 2023 | 807,584 | 5.59 | 23.5% | 12.9% | 11.78% |
| 2024 | 952,701 | 7.01 | 20.6% | 19.4% | 10.17% |
| 2025 | 1,188,007 | 11.50 | 18.0% | 22.6% | 9.29% |
| 2026 | 956,794 | 9.51 | 17.8% | 25.0% | 6.64% |
The share of mods published into a swarm hour has gone from 13.7% to 25.0% — nearly doubled. The share shipping genuinely alone has fallen from 26.2% to 17.8%. Mean rivals per publishing hour peaked at 11.50 in 2025, and 2026 is running at 9.51 across eight months rather than a full year, so the partial-year figure understates the comparison.
This is a supply-side trend, and it lines up with the platform's headline decline. The 1,000-subscriber clear rate has fallen from 13.42% to 6.64% across the same period. Some of that is total volume. Some of it is that a mod's first hour is now contested by roughly four times as many neighbours as it was in 2022.
Crowding is the first thing on this list you can actually move
The reason this matters more than the hour is that crowding is the only variable in this whole body of work that a creator can influence at all.
You cannot change your Steam account's age. You cannot change the year your game's Workshop opened. But you can look at your game's recent publish volume and choose an hour that is not already spoken for — and the quiet slots are not scarce in most games. Cohort-wide, 46.2% of mods already publish into a quiet game-hour, meaning the majority of the catalog is doing this by accident. Even in the busiest of them, quieter windows exist:
| Game | Mods in cohort | Mean rivals per publishing hour | Median | % published into a quiet hour |
|---|---|---|---|---|
| Trailmakers | 326,419 | 19.19 | 17 | 3.5% |
| People Playground | 128,774 | 12.10 | 10 | 11.4% |
| Human Fall Flat | 210,057 | 10.38 | 9 | 11.5% |
| Counter-Strike 2 | 143,149 | 9.92 | 8 | 15.6% |
| Scrap Mechanic | 180,498 | 9.45 | 7 | 18.5% |
| Cities: Skylines | 109,937 | 9.36 | 5 | 33.8% |
| Space Engineers | 150,532 | 8.86 | 6 | 20.1% |
| Brick Rigs | 139,621 | 7.99 | 6 | 24.4% |
| Stormworks | 162,933 | 7.98 | 6 | 20.1% |
| Portal 2 | 114,968 | 5.90 | 5 | 31.6% |
| Left 4 Dead 2 | 114,871 | 5.42 | 4 | 39.4% |
| Garry's Mod | 115,902 | 4.81 | 4 | 38.1% |
| Terraria | 112,797 | 4.62 | 4 | 41.4% |
Trailmakers is the extreme case on the platform: a mean of 19.19 rivals per publishing hour and a median of 17, with only 3.5% of its mods landing in a quiet window. It is also the game with the lowest clear rate in the whole top-volume group, at 0.62% in the morning window and 0.70% in the evening. The crowding is not incidental to that.
The per-game comparison is consistent in the direction that matters. Across the 105 games with at least 100 mods in both a quiet and a crowded cell, quiet hours beat crowded ones in 68 and lose in 34. Restrict to the 13 largest cohort games and the quiet rate wins 12 of 13 — the sole exception is Terraria, and there it is a rounding wash at 3.36% against 3.58%. Cities: Skylines also comes out effectively flat at 13.69% against 13.66%.
What we are not claiming
This is not the publishing-hour argument, resurfacing. We reach the opposite conclusion about the hour. The pooled hour effect is real, large, and clears every conventional control — and it fails the one test that would make it actionable. If the hour were a lever, moving your own release within your own game should move your own odds. It does not: 48.4% of paired creator-game contests go the morning's way, and 51.1% at the game level. The pooled number was reading who publishes when, not what the clock does.
Crowding is not a clean causal lever either. The paired same-creator-same-game test is the strongest control available here and it passes, but a creator choosing a quiet hour is still a creator making a choice. We cannot fully separate "shipped into a quiet hour" from "was organised enough to ship into a quiet hour". What we can say is that the crowd coefficient survives a control the hour coefficient does not.
The mechanism is not identified. Our best reading is competition for a fixed number of front-page slots in a fixed window of player attention: a game-hour absorbs only so much new supply before the marginal mod's share of it collapses. We cannot rule out that crowded hours coincide with game-update days when a flood of low-effort derivative content ships. Tested directly: on the busiest 10% of publishing days per game the clear rate is 8.59%, below the 9.98% cohort average — so burst days are not where the winners hide. And restricting to each game's non-burst days entirely, the crowding gap is untouched: 12.85% in quiet hours against 3.24% in crowded ones, over 3,410,383 mods. That argues for competition rather than event composition. It does not settle it.
One structural exception, stated plainly. Aim-trainer and benchmark games — Aimlabs and KovaaK's — are consumed inside the game, so the Workshop page is not their distribution channel. Any listing or timing property we measure on this platform behaves differently or not at all there. KovaaK's sits at 0.00% clear rate across 43,088 cohort mods, which is the tell. We have flagged the same exception in every piece in this series and it applies here too.
What it means for someone shipping a mod this month
-
Check the hour before you pick it — but not for the reason you were told. The advice you have heard is half right. Picking a quiet hour is worth something. The thing you are picking is the absence of competitors, not the position of the sun. Look at what has shipped into your game over the last week and choose an hour that is empty.
-
The gap is not the biggest one on the platform. 3.3x pooled, and a much narrower 55/45 inside the same creator. If you are choosing how to spend an hour of effort, a preview image and a real description still separate winners from losers by more than this does. Crowding is the lever you reach for after the listing is finished — it is the first one in this series that costs you nothing but timing.
-
The field is getting denser every year, and it is not slowing. If 25.0% of 2026's supply is landing in swarm hours against 13.7% in 2022, then the quiet slots are being consumed. The window that exists in your game today is less likely to exist in it next year.
Publishing at 6 a.m. is not a strategy, and the data says so plainly. But publishing into an empty hour is a real, measurable, and shrinking advantage — and it is the only one in this entire body of work that a creator can pick up for free.