Every paid game on Steam with at least one review, as one dot. Time runs left to right; estimated lifetime earnings run bottom to top on a logarithmic scale. Colour is the game's review score. Hover for a name, click to open its page.

The vertical axis is logarithmic: each gridline is ten times the one below it. On a linear scale the same chart is a flat line and about two hundred visible dots, which is the whole story of Steam's economy in one picture — the button above will show you.

How concentrated is the money? The curve below plots the share of games against the share of revenue they account for. The straight diagonal is what perfect equality would look like.

Worth noting a limitation of the reviews → revenue method: 7,894 paid games have no reviews at all. Too few sales to measure, not necessarily zero sales.

Read this as: a game at the 90th percentile earned more than 90% of all paid games on Steam.

Outcome tiers

Every paid game, sorted into earnings bands. The counts shift when you change the Boxleiter multiplier (see way above), as the bands are dollar amounts and the estimates move underneath them.

What $10,000 in estimated gross actually means for a developer: Valve takes 30%, VAT and regional taxes come off the top in most territories, and refunds and chargebacks come off after that. The figure that reaches a bank account is smaller than the one on this page.

Steam's catalogue grew faster than its revenue. These three charts show what that did to an individual game's prospects.

Games released per year

Median lifetime earnings by release year

The above is by stated release year, and not necessarily the year a game was released on Steam. This one starts at 2003: the handful of games with earlier release years are mostly all-time hits, and their medians are large enough to flatten every year since into the axis. Even from 2003 the fall is steep enough to squash the recent years flat — the button above switches to a logarithmic scale, where each gridline is ten times the one below it, to read them. Recent years are marked “so far”: a game released this year has had less time to accumulate reviews than one released a decade ago, so those bars will keep rising.

Outcome mix by release year

Each year as 100%, split across the earnings tiers. This is the answer to “were a new game's odds better in 2016 than in 2025?” Click any segment to browse the games in it.

Median and top-decile earnings for the biggest genres. A game counts towards a genre only when the tag is one of its six most popular tags — so “roguelike economics” here describes games that are roguelikes, not games that merely carry the tag somewhere down the list.

Hit rate is the share of a genre's games clearing $100,000 in estimated lifetime earnings. Genre names below link through to that genre's shelf on this site.

The same 30 genres, but year by year. Each cell is that genre's median earnings for games released that year, divided by the median for every game released that year. Above 1.0 the genre beat the market; below it, it lost.

Dividing by the same year's market makes this (potentially) useful across time. A ratio compares a genre against its own contemporaries, and both halves have had the same number of years to gather reviews, so the accrual bias divides out.

Rows are sorted by their most recent year, so what's working now sits at the top. Blank cells are genre-years with fewer than 20 games — too few for a median to describe anything. Genre membership uses the same top-six-tags rule as the section above.

Crowding vs. payoff

The same ratio one year at a time, against how many games shipped into that genre. Press play on the timeline below the chart to watch the years run.

Each bubble is one genre in the selected year: how many games shipped, against how they did versus the market. Both axes are logarithmic. Bubble size also tracks release count, so it restates the horizontal axis rather than adding a third measure.

What a game charges, and what it earns. Bands are half-open and aligned to Steam's .99 price points, so nothing is counted twice.

Median earnings by launch price band

Share of all estimated revenue by price band

Free games are counted here but never estimated — there's no price to multiply. See What We Can't Count.

Median launch price over time

Every priced game released that year, not just the top sellers — which is why this line sits well below the figures you'll see quoted for the charts.

The twenty-five highest estimates on Steam. Estimates are least reliable at this end of the curve: review rates vary a great deal between mega-hits, and a percentage point of drift here is worth hundreds of millions of dollars.

Top ten by year

This method of revenue estimation has blind spots! They're listed here rather than in a footnote, because a number is only useful if you know what it excludes.

How the estimate works

Valve doesn't publish sales figures. It does publish review counts, and a game's reviews are roughly proportional to its sales, so multiplying reviews by a constant gives an estimated unit count. Multiply that by price and you have estimated gross revenue.

The approach is named for Mike Boxleiter, who first published the ratio from indie sales data. It has been revisited repeatedly since — Simon Carless at GameDiscoverCo and Jake Birkett among others have published refinements, and the credible range of the constant runs from roughly 20× to 60× depending on genre, price, era and audience. Cheap games bought impulsively tend to review at different rates from expensive ones bought deliberately.

We publish at 35× by default and let you move it, because a single number presented without a dial implies a precision this method doesn't have. Every dollar figure on this page is an estimate with a wide error bar, and it is most reliable in aggregate and least reliable for any single game.

What's counted

• 136,097 released games make up the catalogue we measure — non-game software, adult-only titles and unreleased entries are excluded.
• 99,181 of those have both a price and at least one review, and so receive an estimate.
• 18,331 free games and 10,691 games with no price on record are counted but never estimated.
• 7,894 paid games have zero reviews and sit below the measurement floor.
• 363 listings priced above $120.0 are excluded. Above that line the catalogue is professional software and joke-priced listings rather than games, and a $199 listing with a handful of reviews would otherwise compute to a six-figure result.

Data refreshes continuously; the timestamp at the top of the page is the moment this build was made. Release dates before 1997 or in the future are treated as data errors and excluded.

Cite this page