How Much Do Steam Games Make? The Real Revenue Distribution

Last updated: 2026-06-30 · Reading time: 8 min

Short answer: the median Steam game earns low single-digit thousands of dollars in lifetime gross — far less than the average, because Steam revenue is a power-law (Pareto) distribution where a small top fraction of titles captures most of the money. “Average” is the wrong statistic. After Steam’s 30% cut, refunds, regional pricing, and discounts, developer take-home is well under even that gross headline. None of these catalog numbers can plan your game — forecast it specifically.

Why “average” is the wrong question

When people ask “how much does the average Steam game make,” they usually want the mean — total catalog revenue divided by number of games. That number is real, but it’s nearly useless, because Steam revenue isn’t normally distributed. It’s a power-law: a handful of breakout titles earn millions to tens of millions, while the vast majority earn very little. The few giants drag the mean far above what a typical game actually sees.

Public Steam catalog analyses — VG Insights and Game Discover Co — consistently find that the top ~1-5% of titles capture the majority of all catalog revenue. When a small slice owns most of the pie, the mean describes the slice, not the typical game. The honest summary statistic is the median: the middle game, unaffected by the outliers above it.

And the median is sobering. Across those same public analyses (figures vary by year and methodology), median lifetime gross revenue for a launched Steam game sits in the low single-digit thousands of dollars, with a large share of titles never clearing a few thousand dollars across their entire lifetime. The mean can be several multiples higher — same catalog, very different story depending on which number you quote.

The revenue tiers (lifetime gross)

Here’s the distribution as tiers rather than a single number, with the rough lifetime-gross dollar bands and the wishlist counts that tend to correlate. Wishlist tiers are tied to the floor / mid / breakthrough thresholds from the wishlists-before-launch guide.

Bands are rough P25-P75-style envelopes, not the full distribution, and represent gross (pre-Steam-cut) lifetime revenue. Wishlist correlations are directional — genre, price, and wishlist velocity move them. Your specific game may fall outside any band.

Gross vs net: what you actually keep

Every number above is gross — the headline figure before anyone takes their share. Developer take-home is meaningfully lower, and the gap surprises first-time launchers:

• Steam’s cut: 30%. Valve takes 30% of gross on most sales. The cut steps down to 25% above $10M and 20% above $50M in lifetime gross — thresholds the overwhelming majority of indies never reach, so plan on 30%.
• Refunds. Steam’s no-questions refund window (under 2 hours played, under 14 days) claws back a slice of gross, higher for short or buggy games.
• Regional pricing. A meaningful share of units sell at heavily discounted regional prices (CIS, SEA, LATAM), so unit count overstates dollar revenue.
• Launch and seasonal discounts. Launch-week discounts and Steam sales trade margin for volume; much of your tail revenue is discounted revenue.
• Taxes & splits. VAT/sales tax, withholding, and any publisher or revenue-share agreement come off the top of what reaches you.

Stack those and developer net often lands well under the gross headline. For a deeper look at how much a single wishlist is really worth once all of this is netted out, see what is a Steam wishlist worth.

Why the median is so low

The median isn’t low because most games are bad. It’s low because most games never get seen. Steam’s discoverability surfaces — Popular Upcoming, Recently Released, the recommendation queue — concentrate attention on titles that already have momentum (wishlist mass, early sales velocity). A game below the discoverability floor depends almost entirely on its own channels (social, Discord, newsletter) plus search-query clicks, and those don’t scale to the long tail of impressions that drives real revenue.

The result is a self-reinforcing concentration: the algorithm gives more attention to games that are already converting, those games earn more, and the median game — which never cleared the floor — sits at the bottom. The annual Steam release count has climbed substantially in recent years, which only sharpens the effect: more titles competing for the same finite algorithmic slots pushes the typical outcome lower, not higher.

The punchline: averages can’t plan your launch

Catalog averages — mean or median — describe a population. They’re fine for a journalist’s headline and useless for a founder’s budget. The power-law shape guarantees that any single summary number is wrong for almost every individual game: most titles fall well below the mean, and the median understates the games that did the homework on wishlists and positioning.

Plan against a calibrated forecast of your game instead — and budget against the downside (P10) of that forecast, not the median. A median estimate is right about half the time and badly wrong the other half; the P10 floor is the number you could still survive if everything went sideways. For why single-number heuristics fail and what a probability cone replaces them with, see the Steam launch revenue estimator walk-through.

Frequently asked questions

Built by Greg C. — senior software engineer with production ML experience in calibrated prediction. Steam Launch Forecaster trains a CQR-calibrated model on a 77K-app Steam corpus. See the methodology →