Somebody in your Discord is about to spend two years building a metroidvania. They have not looked up how many metroidvanias shipped last year, or how many of them anybody played. They picked the genre because they love the genre, which is a real reason, just not a business one.
This is the study to read before that. Not to talk you out of your idea: to make sure you are choosing your difficulty setting on purpose rather than by accident.
The one number that reframes everything
In 2025, 20,282 games released on Steam. 608 reached 1,000 reviews. That is a 2.99% success rate, and it is, cheerfully, the highest of the last four years.
A thousand reviews is the industry's least-bad proxy for "this worked". It is not arbitrary vanity: games that clear it almost always clear $150,000 in gross revenue, which is roughly $105,000 after Valve's cut, before tax and before you pay anyone. Chris Zukowski at How To Market A Game has tracked this benchmark for years and it is the closest thing the industry has to an agreed pass mark.
So: 3% is the coin flip you are actually taking. Every genre below is measured against that line. Anything above 3% means the genre is doing work for you. Anything below it means your genre is charging you rent.
How to read the table
Chance of reaching 1,000 reviews, Steam 2025
(scale: one block ≈ 0.7 percentage points)
Job simulator ██████████████████████████████████████████████████ 34.7%
Open world survival ██████████████████████████████ 20.8%
Farming ████████████ 8.3%
City builder █████████ 6.4%
Roguelike deckbuilder ███████ 5.1%
Simulation (all) ██████ 4.1%
Metroidvania ██████ 4.0%
Management █████ 3.4%
Horror █████ 3.2%
─── ALL OF STEAM ─── ████ 2.99%
Idle / incremental ████ 2.8%
Tower defense ████ 2.6%
RPG ███ 2.4%
Puzzle platformer ██ 1.5%
3D platformer ██ 1.5%
2D platformer ░ 0.25%
Two honest caveats before anybody screenshots that. First, these come from two separate passes by the same analyst and a couple of figures differ slightly between them (roguelike deckbuilders read 5.1% in one and 4.87% in the other). Treat them as accurate to roughly half a point, not to two decimals. Second, the 2D platformer figure is from 2024, because the number is so small that a single year of data barely resolves it. That is itself the finding.
Volume is not the same as opportunity
The genres that produce the most successes are not the genres with the best odds, because the popular ones are also the crowded ones.
| Genre | Released 2025 | Hit 1,000 reviews | Success rate | Year on year |
|---|---|---|---|---|
| Open world survival craft | 72 | 15 | 20.8% | down from 24.5% |
| Farming | 60 | 5 | 8.3% | down from 20.8% |
| Roguelike deckbuilder | 212 | 11 | 5.1% | down from 6.7% |
| Simulation | 1,048 | 43 | 4.1% | up from 3.8% |
| Management | 549 | 19 | 3.4% | down from 5.4% |
| Horror | 1,208 | 39 | 3.2% | up from 1.8% |
| Idle / incremental | 965 | 27 | 2.8% | down from 3.1% |
| RPG | 1,158 | 28 | 2.4% | up from 1.5% |
| Adult | 1,846 | 21 | 1.1% | down from 1.3% |
Look at survival craft: 72 games released all year. Seventy-two. Compare that to 1,208 horror games. The genre with the best odds on Steam is the one almost nobody attempts, because it is the one that takes three years and a team.
That is the trade the whole study comes down to. Good odds are expensive. Cheap genres are crowded. There is no genre that is both easy to enter and easy to win, and any advice telling you otherwise is selling a course.
The eight genres worth arguing about
1. Open world survival craft
Odds: 20.8%. The best on the board, and the hardest to reach.
Who you are up against: Valheim, Palworld, Enshrouded, V Rising, Subnautica 2. That last one carried 3.8 million wishlists into 2026, which is the sort of number that tells you where player appetite sits.
| Strengths | Weaknesses |
|---|---|
| Best success rate of any measured genre. Enormous playtime per player, which drives reviews, which drives the algorithm. Co-op sells itself: every player recruits friends. Early Access is genuinely accepted here rather than tolerated. | Multi-year scope minimum. Networking, persistence, procedural generation and inventory systems are each a project. Solo development is close to impossible. Live-ops expectations after launch never stop. |
| Opportunities | Threats |
| Only 72 releases a year means the field is genuinely thin. Setting is wide open: almost everything is fantasy, forest or zombies. A distinctive world is a real wedge. | The bar is set by games with 30-person teams. Player expectations are calibrated to Valheim. The rate already slipped 3.7 points in one year as more entrants arrived. |
Verdict: the highest-EV genre on Steam if you have a funded team and thirty months. Actively dangerous for a solo developer, because the failure mode is not a small game that flops, it is an unfinished game that never ships.
2. Simulation, especially job and shop sims
Odds: 4.1% overall, but 34.7% for job simulators specifically.
That job sim number is the single wildest figure in this study: roughly one in three ships successfully. Competitors: PowerWash Simulator, Supermarket Simulator, House Flipper, Schedule I.
| Strengths | Weaknesses |
|---|---|
| Loop is simple to build and simple to explain. Enormously streamable and clippable. A trailer sells it in eight seconds with no context. Short development cycles, often under a year. | Genre reads as disposable. Little player loyalty between titles. Prices sit low, often $9.99 to $14.99. The premise carries the game, so a weak premise is unrecoverable. |
| Opportunities | Threats |
| Every unautomated real-world job is an unclaimed premise. Rising success rate, one of the few genres improving. | The most obvious jobs are taken and clone velocity is brutal: a hit premise gets copied within weeks. That 34.7% is a snapshot of a bubble, not a permanent condition. |
Verdict: the best odds-to-effort ratio available right now. Ship fast, because you are racing the trend rather than riding it.
3. Horror
Odds: 3.2%, and it nearly doubled in a year.
Horror produced 39 hits from 1,208 releases and has been the top-selling genre on Steam for three consecutive years. Competitors: Phasmophobia, Lethal Company, Content Warning, R.E.P.O. And the outlier that should be on every developer's whiteboard: Cloverpit sold 750,000 copies in two weeks after about a year of development, while Slots & Daggers moved 100,000 units in ten days off eight months of work.
| Strengths | Weaknesses |
|---|---|
| Cheapest genre to make look good: darkness is free, and low fidelity reads as style. Co-op horror is the strongest streaming multiplier on the platform. Short games are acceptable and even preferred. Success rate rising fast. | 1,208 competitors a year. Novelty decays hard, and a scare only works once. Long-tail sales are weak because horror is consumed and discarded. |
| Opportunities | Threats |
| Hybrid horror is where the growth is: horror plus casino, horror plus simulation, horror plus co-op. Nearly half of the top co-op games have a horror element. | The most crowded genre in this study by absolute volume. Discovery is winner-takes-most: a handful of streamed titles absorb the entire month's attention. |
Verdict: the right genre for a small team that needs to ship something this year. Take the hybrid route. Pure atmospheric horror is a much harder sell than horror crossed with a hook.
4. Roguelikes and roguelike deckbuilders
Odds: 5.1%, or 70% better than the platform average.
Competitors: Slay the Spire 2, Balatro, Monster Train 2, Hades 2. Note that deckbuilders are the genre everyone calls oversaturated, and they still beat metroidvanias, tower defense, RPGs and every platformer variant. The saturation complaint is not supported by the data.
| Strengths | Weaknesses |
|---|---|
| Systemic depth substitutes for content volume: build once, replay forever. Enormous playtime per dollar, which reads well in reviews. Genuinely achievable solo. Content updates extend the tail for years. | Balance work is endless and mostly invisible. The genre's ceiling is set by Balatro, an unusually cruel comparison. Hard to demo, because the appeal is on run twenty and not run one. |
| Opportunities | Threats |
| 212 deckbuilders a year is far fewer than the discourse implies. The audience actively seeks new entries rather than sticking to one. | Reputational saturation is real even if statistical saturation is not: "another deckbuilder" is a headwind with press and streamers regardless of what the numbers say. |
Verdict: the strongest realistic pick for a solo developer or a pair. Ignore the saturation talk and bring one genuinely new mechanic, not a new theme.
5. Management and city builders
Odds: 3.4% management, 6.4% city builders.
Competitors: RimWorld, the Two Point series, Cities: Skylines II, Manor Lords.
| Strengths | Weaknesses |
|---|---|
| Fiercely loyal audience with hundreds of hours per player. Mod ecosystems keep games alive for a decade. Steady, predictable long-tail revenue. | Deep systems work with little visual payoff for the effort. UI and UX cost is severe and consistently underestimated. Demos convert badly because the fun starts an hour in. |
| Opportunities | Threats |
| City builders at 6.4% are more than twice the platform average and unusually under-discussed. Colony sims outside medieval and space settings are wide open. | Management dropped from 5.4% to 3.4% in a year, the sharpest fall of any genre here. RimWorld is a ten-year moving target. |
Verdict: solid if you genuinely enjoy systems design, because you will be doing it for a long time with nothing pretty to show. City builder specifically is the underrated pick of this whole study.
6. RPGs
Odds: 2.4%, below the platform average, and rising.
Competitors: Baldur's Gate 3 and Clair Obscur: Expedition 33 have both reset what players expect from a mid-budget RPG, which is a problem for everyone below them.
| Strengths | Weaknesses |
|---|---|
| Highest emotional attachment of any genre. Supports premium pricing, $24.99 and up. Strong review scores when the writing lands. Rate improving, 1.5% to 2.4%. | Content is linear cost: every hour of game is an hour of work, with no systemic multiplier. 1,158 competitors. Writing, VO and localisation stack up fast. Solo RPGs take five years and usually take longer. |
| Opportunities | Threats |
| Narrative was the single largest hit category of 2025 at 51 successes. Short, sharp 6-hour RPGs are underserved while everyone chases 40 hours. | Roughly 19 of those 51 narrative hits were translated FMV titles from Chinese studios, so the headline is less encouraging for a western text RPG than it looks. Player expectations are anchored to $100m productions. |
Verdict: only if the story is genuinely the reason you are making a game. Build for six hours, not sixty. Do not attempt this solo unless you have accepted it as a decade of your life.
7. Idle and incremental
Odds: 2.8%, marginally below average, with 965 releases a year.
| Strengths | Weaknesses |
|---|---|
| Cheapest genre to build by a wide margin. Numbers going up needs no art budget. Very fast iteration. Genuinely a weekend-to-three-months genre. | Lowest revenue ceiling here. Price resistance above $9.99 is severe. Free web and mobile alternatives are infinite and excellent. |
| Opportunities | Threats |
| Named as one of the genres in a current boom. Crosses well with everything: idle plus horror, idle plus management. | 965 releases into a genre players mostly consume for free. Success here rarely funds the next project. |
Verdict: a good first shipped game and a bad career plan. Use it to learn the release process end to end with something small enough to actually finish.
8. Platformers and metroidvanias
Odds: 1.5% for 3D and puzzle platformers, 0.25% for 2D, 4.0% for metroidvania.
This is the section people will argue with, so let me be exact. A 2D platformer had roughly a 1 in 400 chance of clearing the bar. That is not a difficult genre, it is a genre that has stopped functioning as a commercial proposition on Steam.
| Strengths | Weaknesses |
|---|---|
| Easiest genre to prototype and to learn on. Metroidvania at 4.0% quietly outperforms the platform average. Feel and polish are achievable solo. | The worst odds measured anywhere in this study. Genre reads as "first game" to press and streamers regardless of quality. Almost impossible to communicate differentiation in a capsule image. |
| Opportunities | Threats |
| Metroidvania's exploration hook gives it something a pure platformer lacks, which is exactly why it scores 16 times higher. | Silksong. Every metroidvania released this decade is measured against it and the comparison is not survivable. Free platformers are abundant and often excellent. |
Verdict: make one to learn. Do not make one to earn. If you love the movement, build a metroidvania instead: same skills, sixteen times the odds.
What the money actually looks like
Global games revenue reaches roughly $205bn in 2026 on Newzoo's software-only figure, up from $188.8bn. PC specifically generated $39.9bn in 2025 and is forecast to overtake console by 2028, growing at 6.6% against console's 4.4%.
Then it narrows fast. Steam took $17.7bn in 2025, of which indie games were $4.4bn, about 25%. That sounds enormous until you divide it.
| Outcome | Roughly how many games (of 20,282) | What it means |
|---|---|---|
| Over $1m gross | ~300 | 1.5%. A studio, funded for years. |
| Over $150k gross (1,000 reviews) | 608 | 3.0%. One developer, one year, no employees. |
| Under 10 reviews | close to half | Effectively invisible. |
And the part nobody wants on a slide: "Triple-I" studios with near-AAA budgets took 53% of all indie revenue in 2024. The indie market is not $4.4bn shared among small teams. It is roughly half that, and the other half belongs to companies you would not call indie if you met their org chart.
Plan for $150k as a good outcome, not a floor. Anything you build that needs $500k to break even is a venture bet, not a career.
The questions developers actually ask
Is my genre too saturated? Saturation is a ratio, not a headcount. 1,208 horror games with a rising success rate is a healthier market than 60 farming games whose rate collapsed from 20.8% to 8.3% in a single year. Compare rate against the 3% baseline, never raw release counts.
How many wishlists do I need before launch? Treat 5,000 as the floor for a commercially serious release and aim for 7,000 or more. The median indie converts about 12% of wishlists into day-one sales, so 7,000 wishlists is roughly 840 launch-day units. At $14.99 that is about $12,600 gross before Valve's cut. Do that arithmetic before you quit your job, not after.
When should the Steam page go up? Immediately, and long before it is ready. Wishlists never expire, so a page live for eighteen months accumulates a store of value nothing else replicates. There is no advantage whatsoever to waiting.
Does Steam Next Fest work? It is the biggest single wishlist accelerator available, but understand what it does. The strongest predictor of what a game earns during Next Fest is what it had before it, with a correlation of 0.825. The fest amplifies existing momentum, it does not create it. The February 2026 median demo gained 806 wishlists; the 95th percentile gained 13,461. Of games that gained 15,000-plus, nearly all had demos out weeks or months in advance. Shadow-dropping a demo days before the fest is the one pattern that reliably fails.
Do I need a demo? Yes, in almost every genre. Demo-to-wishlist conversion runs around 20%, and a demo is the price of entry to Next Fest. Genres where the fun starts late, like management and roguelikes, need a demo built specifically to compress that: do not just ship your first hour.
What should I charge? $14.99 to $19.99 is the sweet spot for a polished indie. Among the top 1,000 games, $19.99 is the most common price at 147 titles, then $14.99 at 124. Short or viral co-op games work at $7.99 to $9.99. Note the direction of travel: the median launch price for top new releases fell from about $19.50 to $15.64 over three years, driven by viral co-op rather than by a general race to the bottom. Do not discount at launch. Launch revenue determines how much visibility Steam gives you afterwards.
How long should the game be? Shorter than you think, in every genre except survival and management. A tight six hours reviews far better than a padded twenty. Playtime matters for the algorithm, but resented playtime shows up in reviews.
Do I need a publisher? For survival craft and RPGs, funding is usually the deciding factor and a publisher is the realistic route. For horror, sims, roguelikes and idle games, a competent solo developer with a Steam page up early can reach the 3% line unaided. Ask what specifically they do that you cannot, and be unimpressed by "marketing" as an answer.
Hot genre, or the game I actually want to make? The game you want to make, in the most favourable genre that can contain it. Nobody finishes a two-year project they resent, and an unfinished game has a 0% success rate in every genre on the board. Use this study to pick between ideas you would enjoy, not to override the list.
I am solo with no money. What now? Horror, a roguelike, or a job simulator. All three are shippable in under a year by one person, all three sit at or above the platform average, and all three have live examples of small teams breaking through. Put the Steam page up this week.
How soon will I know if it is working? Faster than is comfortable. The Discovery Queue drives nearly half of post-launch page visits and Steam's algorithm largely decides what it thinks of your game by day three. Your launch week is not the start of a slow build. It is the verdict.
What this study cannot tell you
Every number here is backward-looking. The 34.7% on job simulators describes games that shipped into a trend that was already running; by the time your version is ready that window may well have closed, which is precisely what happened to farming games between 2024 and 2025 when the rate fell by more than half in twelve months.
The durable finding is not "make a job sim". It is this: the odds vary by a factor of a hundred between the best and worst genre, and most developers pick without ever looking. A 2D platformer and a survival craft game are both "a game" in a design document and are not remotely the same commercial proposition. Choosing knowingly is most of the advantage available to you.
If you want to see what is actually landing rather than what should, the live board on PokeIndie ranks games by real player activity, and you can filter browse by genre to see who you would be shipping alongside. The horror, roguelike and survival shelves are the three worth studying most closely before you commit.
Then go and make the thing. Just make it with your eyes open.
Written by
MML Tech
Everything by this author →