What does addressable market mean for a Steam game?
It is the number of players who already buy games like yours, and what they pay. It is not the size of Steam. Valve said at GDC 2026 that Steam passed 42 million peak concurrent users and 13.9 million in-game concurrent users. Most of those people will never look at your genre.
A top-down estimate takes a share of a big number, such as 1% of Steam players. Nobody can check that share. A bottom-up estimate adds up what real comparable games sold. You can check every input, and a publisher can repeat your math.
Sources: Game Developer (Chris Kerr, 11 March 2026).
How big is the gap between genres?
Very big, so check your genre before anything else. How To Market A Game counted the 2025 games that reached 1,000 reviews and compared them with all 2025 games that carry the genre tags. 608 of 20,282 games released in 2025 reached 1,000 reviews.
Some genres have few games and a high hit rate. Others have thousands of games and a very low hit rate. The author notes that the totals come from tags, and many developers mis-tag their games, so the totals are approximate.
| Genre | Games with 1,000 reviews | Released in 2025 | Share reaching 1,000 |
|---|---|---|---|
| Open World Survival Craft | 15 | 72 | 20.8% |
| Roguelike Deckbuilder | 11 | 212 | 5.1% |
| Simulation | 43 | 1,048 | 4.1% |
| Horror | 39 | 1,208 | 3.2% |
| RPG | 28 | 1,158 | 2.4% |
| Metroidvania | 3 | 269 | 1.1% |
| Puzzle | 14 | 4,022 | 0.34% |
| 2D Platformer | 3 | 1658 | 0.18% |
Sources: How To Market A Game (Chris Zukowski, 27 January 2026).
How do I build the comp set?
Use the comp method from our market research guide: games in your genre, at your scope and price, from recent years, including the ones that failed. Do not repeat the work for this estimate. Take the same list and add three columns: review total, release year and base price.
Get the review total from Valve's GetAppReviews method. It needs no key. The total_reviews field comes on the first page when the review type is all. Note the release year, because the ratio of sales to reviews changes by year.
Sources: Steamworks IUserReviewsService documentation; GameDiscoverCo (Simon Carless, 3 August 2020).
How do I turn reviews into copies?
Multiply the review count by a sales per review ratio, and use a range. GameDiscoverCo's 2020 survey of 237 games found a median of 58 sales per review across all years, with the middle 80% between about 25 and 100. For games launched in 2020 alone, the median was 38 and the range 20 to 60.
The ratio fell after Steam added a review prompt at the end of October 2019. Older games got many late reviews. One developer reported a lifetime ratio of 30, but 14 for 2020 alone and 67 for 2017 to 2019. So use the ratio for the comp's launch year, and treat old comps with care.
Our arithmetic: a comp with 2,000 reviews and a 2020 range of 20 to 60 sold about 40,000 to 120,000 copies.
| Launch year | Average sales per review | Median sales per review |
|---|---|---|
| Before 2017 | 81 | 74 |
| 2019 | 57 | 51 |
| 2020 | 41 | 38 |
| All years (237 games) | 63 | 58 |
Sources: GameDiscoverCo (Simon Carless, 3 August 2020).
How do I turn the comp set into a market size?
Add the copies of all comps for the total, and take the median comp for your own likely result. The total shows how many purchases the niche made. It is not a count of unique players, because one fan buys many games in the same niche.
Our arithmetic: 10 comps with 30,000 reviews in total, at 20 to 60 sales per review, sold about 600,000 to 1,800,000 copies. If the median comp has 900 reviews, a typical result is about 18,000 to 54,000 copies.
Do not multiply copies by the full price. GameDiscoverCo warns that discounts, the Steam cut, refunds and VAT leave about 30% to 50% of the optimistic gross, and potentially much less if you discount hard.
- Sum copies across all comps: this is the niche size.
- Take the median comp: this is your base case.
- Take the comp at the 75th percentile as an upside case, and the 25th as a downside case.
- Apply 30% to 50% of gross to get a net revenue range.
Sources: GameDiscoverCo (Simon Carless, 3 August 2020).
How do I check demand for the same niche before launch?
Look at the followers of unreleased games in your niche. GameDiscoverCo says follower counts are public for every Steam game. In June 2023, unreleased games had a median of 12x as many wishlists as followers, and most were between 7x and 20x.
The multiplier changes by genre. It was 7.5x for 4X, 9.0x for turn-based strategy and 9.5x for survival games, but 13.2x for story-rich, 15.7x for relaxing and 15.9x for puzzle games. Use the multiplier for your genre to estimate the wishlists of upcoming comps.
Sources: GameDiscoverCo (Simon Carless, 21 June 2023).
What mistakes make a market estimate wrong?
Most errors come from the comp list, not the math. These are the common ones.
- Only hits in the list. A genre with one hit and hundreds of releases is a small market for you.
- Free-to-play comps. GameDiscoverCo removed them as outliers, because free games get many more reviews per sale.
- Old comps with a recent ratio. Games from before 2017 had a median of 74 sales per review, not 38.
- Famous IP or one-language games. How To Market A Game filters these out of its genre counts for the same reason.
- Gross shown as net. Present net revenue, with the 30% to 50% rule stated.
Sources: GameDiscoverCo (Simon Carless, 3 August 2020); How To Market A Game (Chris Zukowski, 27 January 2026).
How do I see which channels reach that market?
Manalens is the analytics and growth engine built for games. Once your Steam page is live, short links and the Steam UTM import show which channels bring visits, wishlists and purchases, so you can compare your real numbers with the comp range. Check single comps with the Steam sales estimator.
Sources: Manalens product documentation.
Frequently asked questions
What is a good number of comps for a market estimate?
Use enough to see a spread, not one hit. Our market research guide suggests about 10 recent games in your genre and scope, including failures. With 10 comps you can take a median and a 25th and 75th percentile case. Fewer comps make the range too wide to plan with.
Can I use the Boxleiter number from older articles?
Use the ratio for the launch year of each comp. GameDiscoverCo found a median of 38 sales per review for 2020 games but 74 for games launched before 2017. Jake Birkett's 2018 survey had a median of 77. A single old ratio overstates sales for recent games.
Is the total Steam audience useful for a pitch?
Only as context. Valve reported 42 million peak concurrent users and 13.9 million in-game concurrents in 2026, but no pitch can claim a share of that. Publishers trust a bottom-up range from named comps more, because they can check each review count themselves.
Should I include games older than three years?
Only with care. Old games had higher sales per review: GameDiscoverCo found a median of 74 for games launched before 2017, against 38 for 2020 games. Old games also sold into a smaller, less crowded store. If you keep one, use the ratio for its launch year and mark it as old.
How do I convert estimated copies to revenue?
Multiply copies by an average paid price, not the full price, because most copies sell on discount. Then GameDiscoverCo suggests that 30% to 50% of the optimistic gross is a realistic net after discounts, the Steam cut, refunds and VAT. Show the range and the rule.
Sources and implementation
- Game Developer: Valve says 5,863 titles earned over $100,000 on Steam in 2025 (Chris Kerr, 11 March 2026) ↗
- How To Market A Game: What the hell happened in 2025? (Chris Zukowski, 27 January 2026) ↗
- Steamworks documentation: IUserReviewsService (Valve, read 18 November 2026) ↗
- GameDiscoverCo: How that game sold on Steam, using the 'NB number' (Simon Carless, 3 August 2020) ↗
- GameDiscoverCo: Why your Steam 'follower to wishlist' multiplier matters (Simon Carless, 21 June 2023) ↗