Guides
How to estimate a Steam game's sales from its reviews
DreamEp, founder of NextWishlist · Updated · 3 min read
On this page
Short answerMultiply the game's Steam review count by a multiplier: GameDiscoverCo's 2020 survey of 237 games found a median of 58 sales per review, 38 for 2020 releases, and practitioners use about 20 to 60 today. Treat the result as a range, and revenue as far below copies × price.
Valve doesn't publish sales, but every Steam game shows its review count, and reviews follow sales. That's the idea behind the Boxleiter method, also called the NB number or review multiplier. Here is how it works, the multipliers public data gives, and why the answer is always a range.
01The method in one line
Estimated copies sold = number of Steam reviews × a multiplier. The multiplier is the average number of buyers per review. Use the total review count on the store page (all languages), not just the positive ones.
The idea comes from developers sharing their own Steamworks numbers: Simon Carless (GameDiscoverCo) and developer Jake Birkett compared review counts with real sales, and the ratio held reasonably well across games.
02Which multiplier to use
- GameDiscoverCo, 2020: 237 games, an average of 63 and a median of 58 sales per review. The middle 80% of games sat between about 25 and 100.
- By launch year: games launched before 2017 had a median of 74, those from 2019 had 51, and those from 2020 had 38. Steam's review prompt, added around October 2019, made players review more, so the ratio fell.
- Today: practitioners work with about 20 to 60, often around 30 for recent releases. That's a working value, not a fresh survey: Steam Page Analyzer describes its 2026 value as a projection calibrated on figures developers disclosed.
Use a low and a high multiplier and keep both: "between 6,000 and 12,000 copies" is an honest answer, "8,700 copies" isn't.
03What moves the multiplier
- Price: cheap games collect fewer reviews per buyer (higher multiplier); expensive ones more.
- Audience: engaged niches such as strategy review more often than casual audiences.
- Age: the ratio has fallen over the years, so a game's release year matters.
- Free weekends, giveaways, bundles and keys: extra owners who rarely review, or reviews without matching revenue.
- Very new games: early buyers review more, so estimates in the first weeks run low.
04From copies to revenue
Don't multiply copies by the list price. Discounts, regional prices, refunds, VAT and Steam's 30% all come off. Carless estimated in 2020 that a developer's net is often 30% to 50% of the optimistic gross, less with heavy discounting. Details in how much Steam takes.
05What it's useful for
- Sizing your genre: estimate the sales of the 10 to 20 closest games released in the last few years. If the median is a few thousand copies, plan your budget on that, not on the hit.
- Pricing: compare what similar games earned at different prices (how to price your indie game).
- Pitching a publisher: show comparable games and their estimated range.
For games that aren't out yet, reviews don't exist: use Steam followers instead, about 10 wishlists per follower. See Steam followers to wishlists.
06Tools that do the math
Data sites such as Gamalytic and SteamDB's owner estimates show estimates for released games; compare them in Steam analytics tools. NextWishlist focuses on what comes before: it follows upcoming games like yours every night, and your own wishlists against them.
How to estimate a Steam game's sales from its reviews: quick answers
How can I see how many copies a Steam game sold?
Valve doesn't publish sales. Estimate them from the review count: multiply the number of reviews by about 20 to 60 sales per review, and keep the result as a range.
What is the Boxleiter method?
An estimate of a Steam game's sales from its number of reviews, multiplied by an average number of sales per review. GameDiscoverCo's 2020 survey found a median of 58, and 38 for 2020 releases.
What review multiplier should I use in 2026?
Practitioners use about 20 to 60 sales per review, often around 30 for recent releases. Use a low and a high value and keep the range.
How accurate are Steam sales estimates?
Rough: developers comparing estimates with their real numbers report errors of 30% to 50% when the multiplier is chosen sensibly, more for free-to-play games, giveaways or bundles.
Sources
Valve's rules can change: the official Steamworks page always wins. Practitioner benchmarks are surveys of developers, given as ranges.
- GameDiscoverCo, “How that game sold on Steam, using the ‘NB number’” (237 games) (August 3, 2020)
- Steam Page Analyzer, “The Boxleiter method explained” (March 3, 2026)
- Valve, “New Revenue Share Tiers and other updates to the Steam Distribution Agreement” (November 30, 2018)
- Valve, Steam Refunds (read Oct 28, 2026)
- GameDiscoverCo, “How Steam followers and wishlists relate” (2025)
The Steam marketing tool for indie developers.
Wishlist tracking, comparison with similar upcoming games, your page and demo checked against Valve's rules, and a daily plan that ties it together. In private beta.