How it works

Most revenue figures on Steam are guesses. Somebody looked at review counts, multiplied by a number, and posted a number. Those numbers float around the internet for years and nobody checks them.

A verified figure is not a guess. It comes straight from Steam, through the read-only sales interface Steam already exposes to Steamworks partners. What you see on a verified page is what Steam reports, day by day.

That is the whole difference. An estimate can be off by a lot in either direction. A verified figure does not drift, does not get stale, and does not depend on anybody remembering the right multiplier.

Every page on gamesELO says which one it is. Verified pages carry real revenue from Steam. Estimated pages carry a range built from public data, and we label them so nobody has to guess.

Steam exposes sales data through a read-only service called IPartnerFinancialsService. To let gamesELO read it, you create a Financial API Group in Steamworks. The key comes from that group, not from your account login.

The group has no users and no apps attached to it. It cannot publish builds, change anything about your games, or move money. It is a viewing window onto sales data you already have access to.

The key covers your whole Steamworks partner account. gamesELO only publishes the specific games you choose to list. Anything you do not list stays private.

Keys are encrypted at rest with AES-256-GCM and never shown again after you submit them. To revoke access, delete the group or regenerate the key on Steam. That takes effect immediately.

Steam lets you limit a key to specific IP addresses. gamesELO runs on Cloudflare’s global network, which has no fixed outbound IP, so leave that whitelist empty or the sync will fail. Deleting the group is your off switch.

Steam reports two figures. Gross is what players paid before anything is deducted. Net is gross minus refunds, chargebacks, and VAT. Net is before Valve takes its cut.

So net is not your payout. Your payout is roughly 70 percent of net under Valve’s standard terms. gamesELO shows that as take-home, and labels it as derived.

gamesELO shows both numbers because they answer different questions. Gross is what buyers spent. Net is what survived the refund flood and the sales tax authorities.

DLC and in-game purchases roll up into the base game. If players bought your soundtrack or a character pack, that money counts toward the base game figure instead of sitting in a separate bucket.

All dates are Pacific Time, matching how Steam reports them. A day boundary that looks a few hours off from your own dashboard is not a bug on either side.

Estimates come from a refined Boxleiter method. Boxleiter is the well-known observation that you can infer units sold from Steam review counts, because reviews come from a fairly consistent slice of buyers.

The formula is simple: estimated units equals review count times a multiplier, and estimated gross revenue equals those units times list price times a realised-price factor.

The multiplier depends on when each review was written. In 2019 Steam started prompting players to review, which roughly halved sales per review. Reviews written in 2019 or later count as 30 sales each; earlier reviews count as 55.

Two adjustments sit on top. Sentiment: games above 90 percent positive keep the base, 80 to 90 percent get 1.25, 65 to 80 percent get 1.6, and anything lower gets 1.1, because mixed games get reviewed less per sale.

Scale: the biggest hits get reviewed more often per sale, so games with 100K+ reviews get 0.7 and 20K to 100K get 0.85. Those numbers come from published sales-per-review research, where 90 percent-plus games sit near 30 and Elden Ring near 20.

The realised-price factor is 0.7. That is the gap between US list price and what buyers actually pay once regional pricing and discounts are averaged in. List price is the ceiling, not the expectation.

For games two years or older, the factor drops to 0.55. Sales that long after launch skew toward deep discounts, sometimes bundles priced low enough that the original list price is fiction by month thirty.

Every estimate is a range, not a point. We publish 0.6 times the midpoint as the low end and 1.6 times the midpoint as the high end. A method this indirect deserves a wide band.

Estimates need a price and at least one review. Free games have no price to estimate from, so they are not estimated or ranked.

Public review data comes as a monthly histogram. That is enough to chart a year of revenue, but not enough to isolate a clean 30-day window, because a calendar month is not 30 days.

For the last 30 days we take the current month so far, then add the matching tail of the previous month, scaled by how much of the current month has elapsed. The result lands close to a real 30-day slice.

For the trailing 12 months we sum the last twelve calendar months of estimated sales, each priced with the factor for that month: 0.7 in the first two years after launch, 0.55 after that.

Each month is priced with the factor that applied at that point in the game’s life. A launch month at full price and a month three years later at 0.55 are not the same sale, and we do not pretend they are.

The monthly histogram omits some reviews that the headline total includes, so we scale the chart up to match the headline count. Trailing and 30-day figures are also capped at the game’s estimated lifetime gross.

A valuation is what somebody might pay for the game today. It starts with your trailing 12 months of Steam net revenue, multiplied by 0.7 for your share after Valve’s cut. That number is your take-home for the year.

We then multiply by 2.2. Indie catalog deals trade around 1.7 to 2.4 times annual profit, and for a solo or small catalog the take-home profit and the earnings number are roughly the same thing.

Genre adjusts for longevity. Strategy gets 1.2, Simulation 1.15, RPG 1.0, Adventure 0.95, Action 0.9, Racing 0.9, Sports 0.9, and Casual 0.8. A game with several genres uses the average of its genre factors.

Momentum compares the last six months against the six before them: revenue for verified games, estimated units for estimates. Games with under a year of history get 1.0. Above 150 percent of the prior half, the factor is 1.3. Above 120 percent, 1.15. Between 80 and 120 percent, 1.0. Below 80 percent, 0.85.

Player sentiment uses your Steam review positive rate, but only once a game has at least 50 reviews. 95 percent positive or better gives 1.2, 90 percent gives 1.1, 80 percent gives 1.0, 70 percent gives 0.9, and below that 0.8.

Verified pages show a plus or minus 25 percent spread around the midpoint. Estimated pages use plus or minus 45 percent, because the underlying earnings figure is itself uncertain.

Treat a valuation as a range to argue with, not a number to bank. It is a transparent formula you can check line by line, which is more than most valuation offers give you.

Estimates are estimates. The review multiplier is a population average, so a game whose audience reviews far more or far less than usual will land off the midpoint. The range exists for exactly that reason.

We do not see revenue outside Steam. Sales on itch.io, GOG, Epic, console, or your own website are invisible to every number on gamesELO. A multi-platform game’s page is only one slice of its actual income.

Free-to-play games are not ranked by estimate, because there is no price to multiply. A game’s earnings may also be so front-loaded that a trailing 12-month valuation describes a moment that has already passed.

Verified pages depend on history backfilling. Older games fill in day by day, newest first, in the background, so a page can take minutes or hours to be complete. The page always shows what date it is synced through.

An unreleased game, or one with no sales yet, cannot be verified. Steam has nothing to report, so there is nothing to show. Verify once you have sales history.

Nothing here is financial advice, a promise of sale, or a forecast. If you are making a real transaction, check the actual numbers on your own Steamworks dashboard.

If an estimate for your game is wrong, verification is the fix. Connect a read-only Financial API Group key and the page switches to real Steam revenue, day by day.

Verification also fixes the trailing 12 months and last 30 days, since those come from your sales data rather than from review math. The valuation spread tightens from plus or minus 45 percent to plus or minus 25 percent at the same time.

Verified data takes priority everywhere. If we ever show an estimate for a game whose owner has connected a key, treat it as a bug and tell us.

If you want your verified numbers off gamesELO entirely, re-enter your key on the Remove form. The published figures go away. Revoking the key on Steam stops gamesELO from reading anything in the first place.