How to Estimate Steam Sales From SteamDB (Owners, Reviews, CCU)

To estimate Steam sales from SteamDB, take the game’s total review count and multiply by 30-40, or take its all-time peak concurrent players and multiply by 8-14 to get week-one units. Both numbers sit on every SteamDB app page, and together they’ll land you within roughly 30-50% of the real figure for most paid indie games.

That’s the short answer. The long answer is that SteamDB is the best free research tool for this job and also the most misread one. The “owners” ranges people quote as sales numbers are neither sales nor SteamDB’s own data. This guide covers what each SteamDB figure means, the four estimation methods you can run from it, and a worked example checked against a real game’s disclosed sales. For the platform-agnostic comparison of approaches, see our guide to estimating Steam game sales — this post is specifically about reading SteamDB.

The cheat sheet: lifetime units ~= total reviews x 30-40. Week-one units ~= all-time peak CCU x 8-14. Pre-launch wishlists ~= followers x 12. Owner ranges are a ceiling, never a sales count.

SteamDB is raw Valve data, presented better than Valve presents it. Concurrent player counts come straight from the Steam API — the SteamDB FAQ is explicit that “the number is directly provided by Valve.” Review counts, follower counts, and price history are first-party data too, so you can take them at face value.

Then there’s the owners section. Here’s the part most people miss: SteamDB does not estimate owners. The owner figures on an app page come from third parties — SteamSpy, PlayTracker, VG Insights, and Gamalytic, all credited in SteamDB’s own attribution notes. Each uses a different methodology, which is why the same game can show “500,000-1,000,000 owners” from one source and a tight point estimate from another, side by side.

And those SteamSpy ranges are huge for a reason. SteamSpy worked by sampling public Steam profiles, and in April 2018 Valve made game libraries private by default. The data source collapsed overnight and never fully recovered. A range spanning 2x from low to high end is normal now. On a $15 game, that spread is millions of dollars of ambiguity.

So treat SteamDB as the ingredient shelf, not the finished meal. It hands you four signals — reviews, CCU, followers, owner ranges — and you do the cooking. The four methods below are ranked by how much I trust them.

The reviews method is the Boxleiter method, and it’s the workhorse. Steps:

1. 01Open the game’s SteamDB page and grab the total review count (all languages).
2. 02Multiply by 30-40 for games released in the last few years. That’s your central estimate of lifetime units.
3. 03Run it again at 25x and 45x for a floor and ceiling.

The multiplier matters more than anything else, and it isn’t one number. VG Insights analyzed over 10,000 games and found the ratio depends heavily on release year: games from 2020 cluster around 30x, while games from 2014 and earlier run 70x or higher, because fewer players left reviews back then. The ratio has tightened over time, which means reviews keep getting more reliable as a sales proxy. Our review-to-sales multiplier guide breaks down the current numbers by genre and year, and the Revenue Calculator applies genre-adjusted multipliers automatically so you don’t have to guess.

Why do this on SteamDB instead of the store page? The review graph. SteamDB charts review accumulation over time, and the shape tells you more than the count does. A clean launch spike followed by a steady slope is organic sales. A sudden vertical jump eighteen months after launch is a bundle or a giveaway — thousands of buyers who paid $2 each, now polluting your multiplier. Bundle spikes mean dropping your multiplier toward 20x. The Steam sales calculator guide covers more adjustment cases.

Reach for the CCU method when a game is too new to have meaningful review counts. The formula:

Week-one units ~= all-time peak CCU x 8-14.

This comes from GameDiscoverCo’s May 2025 analysis of the top 50 Steam debuts from March 2025, the best public dataset on this relationship. Their medians: peak CCU converts to week-one sales at 11.4x across all games — 14.1x without pre-orders, 7.8x for pre-ordered titles, since pre-orders fatten the launch-day player count. Day-1 CCU runs hotter, around 22x for non-pre-ordered games.

You’ll still see an older 5-10x rule of thumb quoted in dev forums. Treat it as the conservative floor: current cohort data puts the median above it, though premium-priced and pre-ordered games genuinely do land in that band. For a typical no-pre-order indie launch, 5-10x just runs low.

The caveats are real. GameDiscoverCo themselves flag variance of 50% or more in both directions: viral games keep selling long after their player peak, while heavily pre-ordered fan-first games undershoot. The multiplier estimates week one only — for games more than a few months old, switch to the reviews method. It also breaks completely for free-to-play games and multiplayer titles where playtime decouples from purchases. One practical note: peak CCU lands a median of 2.5 days after launch for no-pre-order games, closer to 4 days with pre-orders, so don’t run this formula on day one and call it final.

SteamDB shows a follower count and growth graph for every game — data the store page mostly hides. Followers are the most useful signal for unreleased competitors, where reviews and CCU don’t exist yet.

The chain works like this. GameDiscoverCo’s deep dive on followers and wishlists found that wishlists run 7-20x the follower count for unreleased games, with a median of 12x. From there, apply standard launch conversion — 10-20% of launch wishlists converting in week one per our wishlist conversion rates guide. Chained together: 5,000 followers, ~60,000 wishlists, ~6,000-12,000 week-one sales. The Wishlist Calculator models this with genre adjustments.

For released games, the followers method degrades badly. Follower counts keep growing after launch, so the follower-to-sales ratio lands anywhere from 2x to 10x+ depending on age and community activity. Use it as a tiebreaker when other methods disagree. Never as your primary. If you’re tracking your own pre-launch numbers, start with our guide on how to get Steam wishlists.

People search for a “SteamDB owners to sales conversion formula” constantly. I see the queries in Search Console. The straight answer: there is no clean conversion formula, because owners are not sales. Owner counts include bundle keys, giveaway copies, and press and curator keys — everyone with the game in their library, including the ones who paid nothing.

The working heuristic, if you need one:

1. 01Take the owner range from the SteamDB page (remembering it’s third-party data).
2. 02Take the midpoint. “500,000-1,000,000” becomes 750,000.
3. 03Check the review graph for bundle spikes. Vertical jumps mean a slice of owners were $1-3 bundle buyers.
4. 04Multiply by 0.7-0.9 — closer to 0.9 with no bundle history. Drop below that range, to 0.5-0.7, for a game that’s been through Humble Bundle more than once.
5. 05Cross-check against reviews x 30-40. If the two estimates disagree by more than 2x, trust the reviews and assume free keys inflated ownership.

This is an observed pattern, not a published standard. That’s exactly why owners belong at the bottom of the method ranking: useful as a ceiling, dangerous as an answer. A game whose reviews imply 80,000 sales but whose owner range says 400,000-800,000 didn’t secretly sell 600,000 copies. It was in three bundles.

No single method deserves your full trust, but they fail in different directions, and that’s exploitable. My triangulation process for any comparable:

1. 01Reviews x 25 and x 45. Floor and ceiling for lifetime units. The spine of the estimate.
2. 02Peak CCU x 8-14 for week-one units, if the launch was recent. A launch-window estimate that matches the early review graph is strong confirmation.
3. 03Owner midpoint x 0.7-0.9 as the ceiling check. If your review-based estimate exceeds the owner range, your multiplier is wrong.
4. 04Hunt for disclosures. Developers leak real numbers in postmortems, GDC talks, and press releases. One confirmed data point recalibrates every estimate you make in that genre.
5. 05Report a range, weight reviews highest. “Probably 150,000-250,000 units” is a usable answer. “187,375 units” is false precision wearing a suit.

When methods diverge wildly, don’t average them — investigate. The usual culprits: bundles, free-to-play mechanics, long Early Access periods, or Game Pass deals.

Two more conversions before units become money. Multiply by average selling price, not list price — discounts and regional pricing typically pull realized revenue 25-35% below the US sticker. Then subtract 10-15% for refunds (the line item our Steam refund rates data shows almost everyone forgets), then Steam’s 30% cut. And before trusting any estimate, know the base rates: our average Steam game sales breakdown shows what the median game actually sells, and the revenue by genre data tells you whether your comparable is typical or an outlier.

Manor Lords is the perfect test case: a city builder that launched into Early Access in April 2024, with a publisher that discloses sales publicly. Its SteamDB page in June 2026 shows roughly 87,700 total reviews, an all-time peak of just over 173,000 concurrent players, and about 559,000 followers.

The ground truth: Hooded Horse announced 1 million copies in the first 24 hours, then 2 million in under three weeks, then 3 million by February 2025.

Score the methods. The CCU median multiplier predicted ~2.0 million units for the launch window; the real number was 2 million in just under three weeks, with 1 million inside the first day. A startlingly clean hit for a one-line formula. The reviews method at 35x gives 3.1 million lifetime — against a confirmed 3 million sixteen months before our reading, with Early Access sales continuing since, that’s reasonable and slightly conservative. The followers method? 559,000 followers against 3M+ sales implies 5-6 units per follower, a ratio you could never have guessed in advance. Last place, as predicted.

One honesty footnote: Hooded Horse’s figures cover all stores (Steam plus GOG and Epic) while SteamDB’s signals are Steam-only, so the effective multipliers here run slightly hot. It doesn’t change the ranking. For how Early Access launches stretch these curves, see our Early Access strategy guide.

If SteamDB hands you ingredients, the estimation platforms sell the cooked dish. How they compare:

How VG Insights’ estimation method works

VG Insights runs an upgraded Boxleiter calculation: instead of one universal multiplier, it applies multipliers derived from its review-ratio study, adjusted by release year and game profile. The limitation is structural to any review-count model — it can’t see discounts, refunds, or in-game purchases, so revenue figures skew high for heavily discounted games.

How Gamalytic’s estimation method works

Gamalytic’s approach dynamically weights four signals per game, and their own testing credits the blend — not any single signal — for landing 77% of estimates within a 30% error margin. Their estimates fold in price history and discounts, though third-party keys and bundles remain blind spots.

The verdict: for one defensible number fast, Gamalytic or VG Insights. To understand why the number is what it is — and catch the bundle spikes that fool automated models — do the math yourself from SteamDB. I do both. The platforms for breadth, SteamDB for depth on the five comparables that actually matter to my decision.

What is the best way to estimate Steam sales from SteamDB in 2026?

Multiply the total review count by 30-40 for lifetime units, then cross-check with peak CCU x 8-14 for the launch window. Reviews are the most reliable single signal and get more accurate every year as review rates tighten. Our guide to estimating Steam game sales compares the alternatives, and the Revenue Calculator runs the math with genre-adjusted multipliers.

How do I calculate Steam game sales from SteamDB owners?

Take the midpoint of the owner range, check the review graph for bundle spikes, and multiply by 0.7-0.9 depending on bundle history. Treat the result as a ceiling, not an estimate — owner counts include free keys, giveaways, and bundle copies that contributed almost no revenue. The Boxleiter method on review counts beats an owners-based number almost every time.

What multiplier should I use to estimate sales from peak concurrent players?

Use 8-14x peak CCU for week-one units: GameDiscoverCo’s 2025 cohort data puts the median at 11.4x, with pre-ordered games near 7.8x and non-pre-ordered games near 14.1x. Expect plus or minus 50% on any individual game. Sanity-check the result against our average Steam game sales data before believing an unusually large number.

Do these estimates account for refunds and discounts?

No. Subtract 10-15% for refunds per our Steam refund rates data, use average selling price rather than list price, then take off Steam’s 30% cut. Skipping those steps inflates a revenue estimate by 40% or more.

Ready to put these formulas to work? Run your comparables through the Revenue Calculator with genre-adjusted multipliers, and read the review-to-sales multiplier guide for the current numbers by genre.

And remember why you’re estimating in the first place: those comparables earned their numbers with store pages that converted. Run your own page through the Steam Page Analyzer for a free breakdown of your capsule, tags, and conversion signals, then browse the Steam Page Leaderboard to see how the games you just estimated present themselves.