Estimate Steam Sales From CCU: Peak Player Multipliers (2026)

To estimate Steam sales from concurrent players, take a game’s all-time peak CCU from SteamDB and multiply by 11.4 for week-one units sold — 14.1x if the game had no pre-orders, 7.8x if it did. Those medians come from GameDiscoverCo’s 2025 study of real sales data, and they carry error bars of 50% or more in both directions.

That is the whole method in one sentence. This post is the deep version: where the multipliers come from, how to pull the right CCU number off SteamDB (there are three, and two of them will mislead you), a worked example scored against a real game’s announced sales, and the situations where CCU beats counting reviews. It’s the companion to our broader guide on how to estimate Steam sales from SteamDB, which covers reviews, followers, and owner ranges alongside CCU. Here I stay on the player-count method.

The cheat sheet: week-one units ~= all-time peak CCU x 11.4 (median). No pre-orders: x 14.1. Pre-ordered: x 7.8. Day-1 peak CCU runs hotter, around x 20. The formula predicts week one only, not lifetime sales.

Every number below converts a peak concurrent player count into estimated week-one units sold. Pick the row that matches your input and the game’s pre-order status:

Why do pre-orders cut the multiplier roughly in half? Because pre-orders front-load the players. A game with 200,000 pre-orders unlocks for everyone at once, so launch-day CCU is fat relative to the sales still to come. A no-pre-order indie builds its player count as word spreads, so the same peak implies more buyers who haven’t peaked yet.

One provenance note before you build a pitch deck on these numbers. Valve publishes concurrent player counts but has never published a sales conversion for them. Every multiplier in that table is reverse-engineered from games whose real sales the analyst could see, and the medians hide a wide spread. Treat them as the center of a range, never as a formula that outputs truth.

The table above comes from one dataset: GameDiscoverCo’s May 2025 analysis of the top 50 Steam debuts from March 2025, cross-referenced against sales data the newsletter has access to through its Plus product. It’s the best public write-up of the CCU-to-sales relationship I’ve found, and honest about its own spread: individual day-1 multipliers in the cohort ranged from 10x (Bleach: Rebirth of Souls, heavily pre-ordered) to 224x (Nubby’s Number Factory, a cheap viral roguelike that kept selling long after launch week).

Two details from that study matter for using the method correctly:

• —Peak CCU arrives days after launch, not on day one. Median time to all-time peak was 2.5 days for no-pre-order games and 4 days for pre-ordered ones. Run the formula on launch night and your input is too small.
• —The multiplier predicts week-one sales, nothing more. The relationship was measured launch-peak against week-one units. Stretching it to lifetime sales is where most people who quote a “CCU method” go wrong.

What about VG Insights, the other name you’ll see attached to Steam sales estimation? Their contribution is the review side, not the CCU side: co-founder Karl Kontus published a study of 10,000+ games’ review-to-sales ratios that modernized the Boxleiter method, finding recent releases cluster around a 30x reviews-to-sales ratio. Their platform’s estimates lean on that review model, adjusted against Steam’s top-seller charts. VG Insights was acquired by Sensor Tower in March 2025, and their original methodology article on vginsights.com now redirects to Sensor Tower’s blog, so the Game Developer write-up is the surviving citable version. If a tool shows you a CCU-flavored estimate today, the underlying multipliers trace back to GameDiscoverCo-style cohort studies or to the platform’s private calibration data, not to anything Valve publishes.

SteamDB’s player data is first-party: per the SteamDB FAQ, concurrent counts come directly from Valve’s API. The catch is that an app page shows three different player numbers, and only one belongs in the formula.

1. 01Open the game’s SteamDB app page (search by name, or take the app ID from the store URL — our app ID guide shows where it lives).
2. 02Find the “Charts” tab. The headline figures are current players, 24-hour peak, and all-time peak. All-time peak is your input.
3. 03Check the date under the all-time peak. This is the step everyone skips. If the peak happened within roughly a week of release, the formula applies. If it happened months later — after a big update or a deep discount — the number no longer represents launch and the week-one multiplier is invalid for it.
4. 04For older launches, scrub the chart back to launch week and read the launch-window peak off the graph instead. SteamDB’s chart lets you zoom to any date range, so you can recover the honest input even when the all-time peak has drifted.
5. 05Check for pre-orders on the Wayback Machine or in the game’s announcement history, then pick your multiplier row: 14.1x without, 7.8x with, 11.4x if you can’t tell.

One more input trap: demo and playtest players sometimes show up under separate app IDs. If a game ran a huge Next Fest demo, make sure the CCU you grabbed belongs to the paid game, not the demo.

PEAK, the co-op climbing game from Aggro Crab and Landfall, is a clean test case because the developers announced real numbers. It launched on June 16, 2025 at $7.99 with a 38% launch discount to $4.95, no pre-orders, and a Steam page that had only existed for a few days. Per Game Developer’s reporting, it sold 100,000 copies in 24 hours and 1 million copies in six days.

Its Steam peak during that first week was just over 100,000 concurrent players; SteamDB logged 102,799 on June 22, six days in. Run the method:

Score it honestly. The 11.4x median lands within about 17% of the announced figure, which for a one-line formula against a viral outlier is a strong result. The no-pre-order multiplier overshoots by roughly 45% — still inside the stated error bars, and directionally useful. If you had been sizing PEAK as a comparable during launch week with nothing but a SteamDB tab open, the method gives you “about a million units,” and that was the truth.

Now the part that proves the week-one restriction. PEAK kept going: 4.5 million copies within a month per GameDiscoverCo, and its all-time peak CCU was reset to 170,759 on August 17, 2025, per SteamDB — 62 days after release. Apply the formula naively to that later peak and you get ~1.9M units, when the game had actually passed 4.5M long before. The formula didn’t fail; it was asked a question it never claimed to answer. CCU multipliers estimate launch windows. Lifetime estimation belongs to the review-to-sales multiplier.

The Boxleiter method is usually the better default, so it’s worth being precise about the cases where CCU wins:

• —The game is days old. Reviews need volume before the multiplier stabilizes; a three-day-old game with 40 reviews gives you a lifetime estimate built on noise. Peak CCU exists from hour one and stabilizes within 2-4 days. For fresh launches, CCU is the only method with a usable input, and it pairs naturally with our first-week sales benchmarks.
• —The game is free-to-play. Reviews-to-sales is meaningless when there are no sales — an F2P game’s review count tells you nothing about revenue, because revenue comes from in-game purchases made by a fraction of players. CCU is the honest public health metric for F2P: it tracks the active population that monetization is drawn from. You still can’t convert it to dollars without knowing the game’s monetization, but for questions like “is this F2P competitor growing or dying,” the CCU chart answers and the review count doesn’t.
• —The review count is polluted. Bundles, giveaways, and key-heavy launches distort review-based estimates in ways that are hard to correct. Launch-window CCU predates all of that.
• —You care about the launch specifically. For a game with a long tail — years of discounts, updates, DLC — lifetime reviews blend all of it together. If your question is “how big was this game’s launch,” the launch-window CCU is the direct measurement, and the review graph only gives you an indirect one.

And the cases where CCU loses: any game more than a few weeks old (use reviews), games with heavy playtime per player like MMOs and survival games (concurrents run high relative to copies sold, so the formula overshoots), and short single-player games where everyone finishes in a weekend (concurrents collapse fast, so a late reading understates the launch).

GameDiscoverCo’s own caveat on their multipliers: results “can regularly be 50% or more different, both up (for viral/longplayed games) and down (for heavily pre-ordered fan-first games).” Within one 50-game cohort, individual day-1 multipliers ran from 10x to 224x. The medians are real, the spread is also real.

Practical rules for living with that spread:

1. 01Always compute a range, never a point. Peak CCU x 8 and x 14 gives you a floor and ceiling for a typical paid game. Reporting “40,000-70,000 week-one units” is defensible; reporting “57,000” is a guess in a lab coat.
2. 02Adjust for replayability. Roguelikes, co-op games, and anything streamers play for weeks convert players to sales above the median, because the player base keeps buying friends in. Short narrative games sit below it.
3. 03Adjust for price. Cheap games run hot (more impulse buyers per concurrent), premium games run cool. PEAK at $4.95 landing above the median fits the pattern.
4. 04Cross-check against a second method. If reviews x 30 and peak CCU x 11.4 point at the same order of magnitude, believe the overlap. If they disagree by 3x, something structural is going on — pre-orders, a bundle, F2P mechanics — and the disagreement is the finding. Our general sales estimation guide covers the triangulation in full.
5. 05Sanity-check against base rates. Before you believe a comparable sold 300,000 units in a week, look at what the average Steam game actually sells. Most numbers that look amazing are wrong before they’re amazing.

Converting units to money adds its own error bars: use average selling price rather than list price, subtract 10-15% for refunds, then Steam’s 30% cut. The Revenue Calculator chains all of that in one pass, so you can go from a peak CCU number to a net revenue range in about a minute.

Run every CCU-based estimate through these checks before it goes in a pitch, a plan, or a Reddit argument:

1. 01Is the peak from launch week? Check the date under SteamDB’s all-time peak. If not, scrub back to launch and read the window peak.
2. 02Did the game have pre-orders? Halves the multiplier, roughly.
3. 03Is it week-one you’re estimating? The formula answers nothing else. Lifetime needs the review method.
4. 04Does a second method agree within ~2x? If not, investigate before averaging.
5. 05Did you state it as a range? Point estimates from an 11.4x median with 50% error bars are theater.

If you’re estimating competitors to size your own launch, the same player math runs in reverse: your wishlist count converts to launch sales at a somewhat predictable rate, and your launch sales imply the CCU that Steam’s algorithm and the charts will display to the world. Model the whole chain with the Revenue Calculator, and if the goal is making your own launch peak bigger, start where every buyer starts: run your store page through the free Steam Page Analyzer and fix what’s leaking before launch week does the measuring for you.