How the Steam Algorithm Works: Discovery and Visibility Explained

Steam is not one algorithm

Most questions about "the Steam algorithm" assume a single ranking system that decides how much exposure a game gets. There is no such system.

An algorithm, here, just means the rule Steam uses to pick which games to put in a given slot. Steam has many of those rules, running in different places, choosing from different pools of games, for different reasons.

The Discovery Queue is a personalized recommendation sequence. The New Releases Queue is a launch-window baseline. New & Trending is a competitive list. Tag pages are eligibility lists. None of them share a ranking.

The distinction changes the diagnosis. "My game has no visibility" is really a question about which specific surface is failing and why, and the answer is different for each.

The surfaces, and what feeds each one

The terminology across those rows is the single most confusing thing about this topic, and the two queues are the worst offenders.

The Discovery Queue is a personalized recommendation feature available at any time, while the New Releases Queue is a launch-window baseline. They are unrelated systems with similar names. Our Steam discovery glossary defines each of these terms in one place.

What Steam counts as a signal, and what it ignores

Valve's public guidance on visibility is unusually direct about its inputs:

Two of those rows do real damage when a studio gets them wrong.

Traffic is not a lever. A studio that buys 10,000 clicks to its Steam page has not bought any Steam-side visibility. If those 10,000 visitors produce 200 purchases and those players keep playing, Steam responds to the 200 purchases and the playtime.

The 10,000 visits are invisible to the ranking, so "drive traffic to the store page to trigger the algorithm" is a strategy that quietly fails: the mechanism it assumes does not exist.

Wishlists are not a general visibility currency. They are worth collecting for two concrete reasons.

First, Steam emails wishlisters when a game launches, when it transitions out of Early Access, and when it is discounted by 20 percent or more; separately, you can send one notification of your own when an associated free demo goes live for public play.

Second, Popular Upcoming is the example Valve names when it says wishlists mostly are not a visibility factor: wishlists are an input there, alongside the release date Valve says such lists depend on. Neither reason is "the algorithm shows my game more because the number went up."

The practical restatement: Steam rewards the response your marketing produces. Everything you do off-Steam has to convert into purchases and playtime before Steam registers it at all.

Visibility changes shape at every stage

The set of surfaces open to you changes at three points in a game's life, and a plan built for one stage does not work in another.

Three specifics inside that table are worth stating plainly, because each one is a decision:

Coming Soon has no baseline. Before release, Steam owes you nothing. There is no queue that automatically shows an unreleased game to anyone. Everything pre-launch is either earned through relative interest, which is Popular Upcoming, or driven by tag relevance.

Coming Soon is the stage where studios most often expect the platform to do work it does not do.

Leaving Early Access is a second launch. Valve treats the 1.0 transition like a full first release for visibility purposes.

Valve is explicit that an Early Access title does not receive its full launch visibility until it is fully released, so nothing is spent on a rough Early Access build: the 1.0 transition is the only full-release baseline the game ever gets.

The timing of that transition is therefore a real marketing decision rather than a production formality.

You start with five Update Visibility Rounds, and they are shared. Five to begin with, shared across the product's life and split between Early Access and full release however you choose.

Each runs up to 30 days or one million home-page impressions, whichever comes first, and each is meant for a major update, DLC or feature. Valve says additional rounds may be granted to products that are selling well, so the allocation can grow, but do not plan around it.

A studio that spends three of them on minor patches in its first year has given up three of its largest post-launch promotional moments, and earning more of them depends on already selling well.

The full stage-by-stage map, including what to check at each handoff, is in which Steam discovery surfaces are open at each launch stage.

How much visibility is at stake

This is the question every studio asks about a placement, and it is worth being straight about the answer: no one outside Valve can quantify it, and Valve does not publish it.

There is no published qualifying number for New & Trending entry, no promised duration once you are there, and no figure anywhere from Valve for how many impressions any placement delivers.

What Valve does say is that a title can be pushed off quickly when other popular products release at the same time, which makes your launch week's competition part of your result rather than a detail.

Percentage claims about how placement traffic is distributed do circulate. Treat them carefully; we are not going to repeat one here.

Impressions for a game you do not own are not obtainable by any public method, so any figure describing how impressions are shared across a genre is either an estimate presented as a measurement, or it comes from a dataset nobody has published.

What follows is a measurement discipline rather than a benchmark: record your own placement window in Traffic Breakdown and compare your game against itself before, during and after it, because there is no cross-game benchmark to use instead.

The practical consequence for planning is that a placement you cannot schedule, whose duration is not promised and whose traffic nobody publishes, is upside rather than a line in the plan.

What New & Trending actually gives you covers what is and is not knowable about that specific placement.

How to see your own visibility, and how far to trust it

Steam does show you your own visibility data, at a level of detail that is genuinely useful, in the Steamworks partner backend: from the product landing page, under Marketing Tools & Data, click Marketing & Visibility, then open the Store & Steam Platform Traffic Breakdown tab.

It reports, per Steam feature:

• Impressions: how many times your game appeared in that placement
• Visits: how many store-page loads that placement produced
• Traffic by source: Discovery Queue, search, tag and browse, More Like This and recommendations, external, Steam community, direct

Read it as a two-step funnel per surface, because the two steps fail for different reasons:

• If impressions fall while the visit rate holds, Steam is showing the game less and the diagnosis is about eligibility, tags or competition.
• If impressions hold while the visit rate falls, the placement still exists and the capsule, the cover image Steam shows for your game in lists, is losing the click: a creative problem.

Two limits to hold onto:

Treat the source split as directional. Valve publishes no accuracy figure for its traffic source categories, but it does document the gaps:

• An impression means only that your game was on screen, which Valve says "doesn't guarantee the player has seen your game".
• Some platform impressions, such as the pop-up telling a player a friend has launched your game, do not link to the store page at all, so click-through rate is not measurable for them.
• Since 23 March 2021, the impression and visit reports exclude players who opted out of applicable cookies.

Treat the breakdown as a reliable shape and an unreliable decimal, and do not build a conclusion that only survives if the external-versus-organic split is precise.

It cannot prove causation. The report is anonymous and placement-level. It will show you that impressions rose the week after a campaign.

The report cannot tell you the campaign caused it, and given that raw traffic is not a visibility input, the honest default assumption is that it did not, at least not directly.

How to read your Steam Traffic Report covers the weekly workflow. If impressions are the thing that is missing, why is my Steam game getting no impressions is the diagnostic sequence.

What you actually control

Most of the inputs to Steam visibility sit downstream of things a studio cannot directly move: you cannot make players buy, schedule a New & Trending placement or raise a rank by spending.

What is genuinely inside your control is a short list:

That last row is the one that changes over a multi-game career. Every other lever in the table is a way of asking Steam for exposure on better terms.

An email list, a Discord, and past buyers you can reach directly are demand you can reach on launch day regardless of what any Steam surface decides. They are also the only assets here that compound across titles.

Steam will not tell you which of your channels produced the players it rewarded you for. Its UTM Analytics credits conversions within a 72-hour window of a tagged click, which is a floor rather than a total, and the Traffic Breakdown's source split is directional.

That measurement gap is a real constraint on optimizing any of the levers above, and it is the subject of marketing attribution for Steam games.

Sources: Valve's Steamworks documentation on Visibility on Steam, Early Access, Steam Tags, Wishlists, Release Dates and Store and Platform Traffic Reporting. Last verified 18 September 2026.