The Steam Workshop's Visitor-Counter Gap (2026): In 80 Ecosystems — Holding 29% of Every Subscription — Mods Report More Subscribers Than Visitors

A Steam Workshop mod cannot have more subscribers than people who have ever looked at it. You cannot install what you never saw.

1,941,933 mods disagree. That is 32.5% of every active Workshop item carrying at least one subscription and a non-zero visitor count in ModSignal's dataset of 7,362,361 live mods.

This is not a rounding error and it is not a tail phenomenon. It is a measurement gap the size of a continent, and it quietly invalidates a whole class of published Workshop statistics — including some of ours.

The impossible number

Steam's Workshop API returns three counters for every item: subscriptions, favorites, and views. The third one is labelled, on the item's own community page, "Unique Visitors". We confirmed that label directly on the live rendered page for every item cited in this article.

A unique visitor is someone who opened the page. A subscriber is someone who installed it. So the ordering is not a matter of opinion:

unique visitors  >=  subscribers
unique visitors  >=  favorites

Here is what the full catalogue actually looks like:

The last row is the cleanest proof available. One thousand six hundred and fifty-nine mods have been bookmarked by real people and have never been viewed by anyone — as far as the counter is concerned.

There is also a fourth counter (votes_up) that passes the test perfectly: zero mods report more upvotes than visitors. So the visitor field is not a corrupt column. It is a column that is sometimes not filled in.

It is not our crawler. Steam says the same thing.

The obvious suspicion is a bug on our side. We checked it against Steam's own API on the mods where the gap is widest.

Every row above was pulled live from ISteamRemoteStorage/GetPublishedFileDetails and from the rendered community page. Steam's numbers match our database to the individual subscriber — 1,470,517 / 162 returned by the API, 1,470,354 / 162 in our snapshot minutes earlier. We are faithfully recording a value Steam is deliberately not filling.

The 2^32−1 problem is Steam's too. One mod in our catalogue reports 4,294,967,295 favorites — exactly 2^32−1, the classic unsigned-integer overflow sentinel. Its community page, served by Steam, shows the same number. That one is a Steam-side overflow, not a collection artefact, which is the useful part: it tells us the counters come with no completeness guarantee at all.

Where the counter is filled — and where it is not

The gap is not scattered. It is organised by game ecosystem.

Across the 392 ecosystems with at least 100 subscribed-and-viewed mods:

• 80 report more subscribers than visitors for half or more of their catalogue
• 22 report it for 90% or more

Those 80 ecosystems are not marginal. They hold 3,637,687,995 subscriptions — 29.2% of the entire platform's 12.45 billion — while contributing just 4.7% of all reported page visits.

What the pattern means

The shape of the list is the explanation. Every ecosystem at the top of that table distributes mods inside the game, not through the community website.

Wallpaper Engine subscribes wallpapers from its own in-app browser. Aimlabs hands you a template picker in-game. House Flipper, Trailmakers, Brick Rigs and Dungeon Alchemist all surface mods through an in-game menu or a built-in storefront. Players never load a steamcommunity.com page — so the unique-visitor counter, which is a web-page counter, records almost nothing while subscriptions flow normally. One Aimlabs mod has 1.47 million installs on 162 page views. Nobody browsed it. 1.47 million people used it.

The ecosystems at the bottom work the other way. Hearts of Iron IV, Skyrim, Stellaris, RimWorld and the Total War series have no in-game mod browser worth the name. Their players research on the website — reading descriptions, checking load orders, comparing alternatives — before subscribing. There the counter is not just complete, it is generous: Skyrim records 5.0 unique visitors per subscription, because browsing is the interface.

Steam's curated/microtransaction lane (file_type = 1) points the same way: 69.3% of those items are inverted, against 32.5% of ordinary community mods. They are bought in-game.

So the honest statement is not "Wallpaper Engine players subscribe to seven wallpapers for every one they preview." It is: Steam's unique-visitor counter measures web-page loads, and in a growing share of the Workshop the web page is not where discovery happens.

We previously published the opposite framing in Steam Workshop Conversion Rates by Game, where Wallpaper Engine's apparent 725% conversion and Project Zomboid's 114% were read as evidence of frictionless in-app adoption. The observation about in-app adoption was right. Calling it a conversion rate was not — you cannot divide installs by a counter that was never collected. That ranking is a map of the counter's blind spot, not of player behaviour.

The corrected ladder: attention per subscriber

The size ladder we set out to measure — how much attention each subscriber actually represents as a mod grows — has to be computed only where the counter is coherent. We defined that set from the data, not by hand: every ecosystem with at least 1,000 subscribed-and-viewed mods and fewer than 10% inverted items. That yields 124 ecosystems.

The decline is monotonic and it is 12.5x end to end: a first-subscriber mod in a browsable ecosystem represents nearly twelve page visits; a million-subscriber mod represents less than one. Past 100,000 subscribers the metric crosses 1.0 — those mods are installed more often than their pages are opened, because by then they travel by word of mouth and collection, not by browsing.

The effect is not an artefact of mixing games. Inside single ecosystems with the largest, most complete counters:

And it survives the two controls that matter. Comparing each ecosystem against itself, the small-mod band beats the big-mod band in 70 of 70 cases. Comparing each creator against themselves — same account, small stack versus million-subscriber stack — the small stack wins in 154 of 154 cases with looser bounds, and 8 of 8 under the strictest ones. Whatever this is, it is not "big games are different" and it is not "big creators are different."

It is the same attention-dilution law we have measured before, but now measured properly, on the only subset of the Workshop where the unit of measurement is real.

What this changes

Three things.

One. Any Workshop statistic that divides subscriptions by views is only valid inside the 124 ecosystems where the counter is filled. For the other 80 — including the single largest ecosystem on the platform by subscriptions — the ratio is undefined, not small.

Two. Comparing mods across ecosystems on views is comparing a web-page metric to an in-game experience. Wallpaper Engine's 1.93 billion subscriptions (15.5% of the platform) arrive through 1.6% of reported page visits. A creator in a browsed ecosystem and a creator in an in-game ecosystem can have identical success and report view counts three orders of magnitude apart.

Three. If you are reading this dataset and your mod's view count looks implausibly low — 162 visitors on 1.47 million installs is a real, live example from Aimlabs — that is not your mod failing. It is the counter not being collected where your players actually are. We report what Steam returns; we cannot make it complete. Where you see a number like that in our data, read the ecosystem column first.

The Workshop's most reliable statistic has always been the one nobody argues about: the install count. Its least reliable is the one everybody divides by. We will keep the visitor counter in the dataset — it is real, and in 124 ecosystems it is the best attention signal available. But we will stop treating it as a platform-wide denominator, and we would suggest everyone else do the same.