How many Steam games are never played?
About a quarter to a third of owned copies, in most public samples. The share is much higher for bundle and giveaway copies, and for players with very large libraries. No sample is a random sample of Steam, so read each number with its method.
GameDiscoverCo looked at public Steam profiles through its partner GamingAnalytics.info. On average, 32.7% of the games in a Steam library were unplayed. The median player had 51.5% of their library unplayed. That second number is the source of the claim that most games go unplayed, and it is a share of a library, not a share of the buyers of one game.
| Source and date | Sample | Finding |
|---|---|---|
| GameDiscoverCo with GamingAnalytics.info, March 2024 | Public Steam profiles | 32.7% of games in the average library unplayed; the median player has 51.5% unplayed |
| imho.run, September 2026 | 113,781 public libraries, paid games | 32.4% of paid copies never launched; 59.1% of likely bundle or giveaway copies against 24.5% of the rest |
| imho.run, September 2026 | Same sample, launched copies | 27.9% of launched copies had less than 2 hours |
| Ars Technica, April 2014 | Public Steam data, games released after March 2009 | 26.1% of registered copies unplayed |
| Ultra survey, October 2023 | 2,003 PC gamers in the US and UK, self-reported | Gamers estimate 24% of their digital games were never played |
Sources: GameDiscoverCo (4 March 2024); imho.run (30 September 2026); Ars Technica (16 April 2014); Ultra (October 2023 survey).
Why do the numbers differ so much?
They measure different things. A share of a library weights collectors who own thousands of games. A share of copies weights popular games. A median player and an average player are also different people.
The samples also differ. imho.run found its libraries by walking Steam friend lists, so its own method note says the sample leans to engaged players with large libraries. It left out libraries where every game shows 0 minutes, because Steam shows that when a player hides playtime. Steam does not say when a game was bought, so a recent purchase looks unplayed. Steam began to record playtime in March 2009, which is why Ars Technica corrected its first figure of 36.9% to 26.1% for games released after that date.
Sources: imho.run (30 September 2026); Ars Technica (16 April 2014).
What is play rate and how do I calculate it?
Play rate is the share of owners who launched the game at least once. Divide the players who launched by the copies owned. Steamworks shows both numbers on the product page of your game.
One public example: the solo developer of Alchemistry shared first-year Steamworks figures in May 2023. The game sold 726 copies, with 101 returns, and had 678 lifetime unique users, which the developer reported as 93%. Its demo had about 14,000 total units and about 3,000 lifetime unique users, which is near 21%.
One game is not a benchmark. It does show that a game bought on its own can have a play rate far above the library averages.
- Open the Steamworks sales reports and choose your game under Products.
- Read the lifetime units and the lifetime unique users.
- Divide lifetime unique users by lifetime units.
- Do the same for your demo app, and keep the two numbers apart.
Sources: Dysnomia, 1 year after launch: figures about Alchemistry (20 May 2023).
How do I measure play rate with Manalens?
Compare first launches from the game with Steam units sold over the same weeks. The Godot and Unity SDKs send game_open with is_first_open true on the first launch of each install. With a Steamworks Financial API key (the automatic Steam sync is in early access), Manalens syncs units sold and returned per day and country.
Be precise about what this is. Manalens does not join a Steam purchase to a player. Steam sales arrive as daily totals per country, and a first launch is an event from one install. The ratio of the two is an estimate for a period, not a list of buyers who never played.
Several things move the ratio. One buyer with two PCs makes two installs. A player who refuses consent, when you start the SDK with consent_required, sends no events. Steam key activations are not units sold. Sales days are Pacific Time and wishlist days are GMT, as Steam reports them, while reports in Manalens use UTC days, so compare weeks, not single days.
- Install the SDK with storefront set to steam, and set enabled to false in the editor.
- Connect the Steamworks Financial API key (the automatic Steam sync is in early access) under Settings, Integrations, Steam, and run the first sync.
- Count first opens per week: game_open with is_first_open true, filtered to Storefront is steam.
- Take units sold minus units returned for the same weeks from the Steam sales data.
- Divide first opens by net units, and watch the trend, not one week.
Sources: Manalens Steam, install attribution and Godot SDK documentation.
Do players who never launch affect refunds and reviews?
They can refund, but they cannot review. Steam accepts refund requests within 14 days of purchase and with less than 2 hours of playtime, so an unplayed copy is always inside the playtime limit for 14 days. Only reviews from accounts that bought the game on Steam and played it count toward the review score.
So a low play rate does not hurt your score directly. It means fewer possible reviewers, fewer players who tell friends, and fewer people in your retention cohorts.
Sources: Steam Refunds policy; Steamworks User Reviews documentation.
What can I do about players who never launch?
First, find out whether it is a launch problem or a library problem. A game that crashes before the SDK starts sends no event, so a crash at startup looks like a buyer who never played. Read the Game health board for app_unclean_exit by build, and check the first launch on each platform yourself.
Then look at the players who do launch. The first session funnel shows how many reach the core loop, and D1 retention shows how many return. Those numbers you can change with a build. Sale weeks and bundle deals bring copies that sit in libraries, as the imho.run bundle numbers show, so compare play rate across sale and full price weeks before you act.
Sources: Manalens error tracking and metrics documentation; imho.run (30 September 2026).
How do I see first launches next to Steam sales?
Manalens is the analytics and growth engine built for games. Add the Unity or Godot SDK to your Steam build and connect a Steamworks Financial API key (the automatic Steam sync is in early access). Then read first opens, units sold and returns by week in one project.
Sources: Manalens product documentation.
Frequently asked questions
Do most Steam buyers never play the game?
Not for most games bought on their own. Public samples put unplayed copies at about a quarter to a third. GameDiscoverCo found the median player has 51.5% of their library unplayed, but that is a share of a library. imho.run found 59.1% of bundle or giveaway copies never launched.
Where do I find my play rate in Steamworks?
Open the sales reports, choose your game under Products, and divide lifetime unique users by lifetime units. Do the same for the demo app separately. The numbers are lifetime totals, so they include early buyers and recent buyers, and a recent buyer may still launch the game.
Can Manalens list the buyers who never launched?
No. Steam does not share buyer identities, and the automatic Steam sync (in early access) gives daily totals per country. Manalens counts first launches from the game. You can compare first launches with net units sold by week, which gives an estimate of play rate for each period.
Do unplayed copies count in my review score?
No. According to the Steamworks documentation, only reviews from accounts that bought the game on Steam and played it count toward the score. Key activations do not count. An unplayed copy can still be refunded within 14 days of purchase, because it has less than 2 hours of playtime.
Sources and implementation
- GameDiscoverCo: How many PC games get bought, but not played? (Simon Carless, 4 March 2024) ↗
- imho.run: Steam Graveyard 2026, the paid games people buy and never play (30 September 2026) ↗
- Ars Technica: Steam Gauge, addressing your questions and concerns (Kyle Orland, 16 April 2014) ↗
- Ultra: PC gaming report white paper (survey of 2,003 PC gamers, fieldwork 13 to 16 October 2023) ↗
- Dysnomia: 1 year after launch, figures about Alchemistry (20 May 2023) ↗
- Steam Refunds policy (Valve, read 18 November 2026) ↗
- Steamworks documentation: User Reviews (Valve, read 18 November 2026) ↗