A data scientist named Ross published I Analysed 62,000 Indie Developers. Here's What Actually Predicts Success, and its message to indie developers is a hopeful one: most of your competitors quit, the odds improve dramatically with every game you ship, and by your fifth game a 25,000-copy hit is "closer to a coin toss". We keep our own copy of the entire Steam catalog around here — 115,281 games, 48,049 developers — so we did what you should always do with a claim that has numbers in it: we ran it. Every claim in the video that can be checked against data got checked, twice — once by me, once by a second model instance that was shown the claims but none of my work, so the two analyses couldn't contaminate each other. The short version: the video's survival numbers replicate almost perfectly, and its success numbers are a survivorship illusion — the odds do not climb because you shipped more games; they climb because the people still shipping are the ones who already succeeded. The long version, with the actual data to poke at, is below.
How any of this is measurable
Steam doesn't publish sales, so everything here runs on the standard proxy: review counts. Players leave roughly one review per 30–60 copies sold; this site's own games give us a rare calibration point, because we know their true sales — both SNKRX and BYTEPATH landed at about 45 copies per review, three years apart, so that's the constant used throughout. The video's success bar is 25,000 copies, which at a mid price is roughly a quarter million dollars gross; at 45 copies per review that's 556 reviews, and that's the definition of "hit" in every chart below. Developers are identified by the developer name on their games, and their releases are ordered by date, which turns the whole catalog into 48,049 little career histories.
Two honest limitations to keep in mind, because they bend several numbers you'll see: the catalog is the live catalog (delisted games are invisible), and old games have had more years to accumulate reviews than new ones. Where a number depends on either of these, the fix is described in place — this post tries not to gloss over the boring parts, because the boring parts are where the video goes wrong.
Claim 1: "5% of games account for nearly 90% of sales"
True — almost exactly. Sort all 97,271 paid games by estimated units and the top 5% hold 87% of all copies. By revenue it's harsher still: the top 5% hold 94%, the top 1% alone hold 80%, and you need only the top 2.8% of games to cover 90% of every dollar. This is what a power-law market means concretely: there is no "typical" game — the median and the mean live in different universes, and averages you read anywhere about Steam income are dominated by a few hundred titles.
The curve below is the whole catalog. The horizontal axis is the share of games (log scale — each step is 10×), the vertical is the share of everything sold. Move your cursor along it; the red dot marks the video's 5% claim.
Claim 2: "Only 1 in 5 developers ever make another game"
The raw number replicates on the nose: 21.7% of developers in the catalog have a second release. But this number has a built-in bias worth understanding, because it's the same bias that will matter in every career statistic below: right-censoring. A developer whose first game came out last year hasn't quit — they just haven't had time to ship a second game yet, and the raw statistic counts them as quitters anyway. The fix is to only look at developers whose first release is old enough for a follow-up to plausibly exist. Do that and the number climbs steadily with the lookback: first releases at least 3 years old — 27%; at least 5 years — 31%; at least 8 years — 38%; at least 10 years — 47%. The honest lifetime statement is that roughly a third to nearly half of developers ship again, given time. "One in five" is what the same reality looks like through a window that's too short.
Claim 3: the survival curve
The video's most striking chart says the chance of quitting is about 60% after your first game and falls to 20–30% once you've shipped two or three. Ours is below, and the censoring lesson applies here in the sharpest possible way, so the chart has three buttons. 3Y WINDOW asks: did a follow-up arrive within a fixed three years? — the fair way to compare everyone, and the harshest: 80% of first games get no follow-up within three years. LIFETIME 8Y+ asks: among developers we've been able to watch for eight-plus years, did a follow-up ever arrive? — and there the curve reads 64% → 35% → 23%, which is nearly exactly the video's claim.
So the video's survival curve is right, as a statement about lifetimes. What it doesn't say is that the same curve viewed on any humanly-relevant horizon is a lot steeper. Both views agree on the shape that matters: the quit hazard collapses with every game shipped — a developer on game five almost never stops.
Claim 4: "1 in 10 on your first game, a coin toss by your fifth"
Here's the centerpiece, and it does not survive contact with the data. The claim: your first game has about a 10% chance of a 25,000-copy hit, and by your fifth game the chance approaches 50%.
The first half checks out at face value — across all eras, 10.1% of first games eventually pass 556 reviews. But even that number is inflated by an artifact worth knowing about: a "first Steam release" before ~2010 is usually an established studio importing its back catalog, and those "first games" hit at 49.8%. Filter to developers who actually started in the Steam era (2015+) and a first game hits at 8.8%; a modern self-published first game is closer to 6%.
The second half is off by a factor of four to five. The chart below shows the real curve for Steam-era developers: from 8.8% on game one it rises to a peak of 11.8% on game three, then declines — game ten sits at 5.8%. There is no climb toward a coin toss. There isn't even a climb. The red line is the video's claim, drawn to scale against reality.
And now press SPLIT BY PAST, because this is the most instructive toggle in the post. It splits each bar by the developer's best previous game: gray = never got traction (best prior under 50 reviews), blue = mid traction (50–555), green = already had a hit. The entire structure of the curve falls out: on game five, a developer whose four earlier games all died sits at 0.2%. One whose best prior was mid-traction: 3.4%. One who already hit: 33.6%. The bar's overall height is just these three groups averaged by how many of each are left — and since the failures quit (see the survival curve above) while the successes keep going, the mix shifts greener with every index. That's survivorship: the odds of the survivors were never the odds of the starters. Shipping more games does not raise your odds. Having succeeded before raises your odds, and the count of games contributes approximately nothing — within every color, the bars actually drift down with experience.
The uncomfortable corollary: veterans with several shipped-but-dead games perform worse than first-timers (partly because serial no-traction shippers include a lot of asset-flip factories — but that is itself information about what "just keep shipping" selects for). And the video's follow-up claim — "less than 15% of new devs ship five games with at least one hit among them" — is technically true but generous: measured, it's 1.8% to 4.6% depending on the starting cohort, mostly because shipping five games at all is rare (4–10%).
The ladder: what actually compounds
If game count doesn't predict success, what does? The single strongest variable in the entire corpus: the best outcome you've achieved so far. The chart below is every developer's next-game hit odds as a function of their best game to date, and it's a smooth staircase with no cliff — from 1.1% (never cleared 10 reviews) through 3.5% (cleared 50), 10.8% (cleared 250), 22.5% (cleared 556 once), up to 67% for developers whose best game passed 5,000 reviews.
Two things to read off it. First, there's no magic threshold — every rung of traction you land raises the next game's odds, including the small ones: getting from "under 10 reviews" to "a couple hundred reviews" is already a 10× improvement in your future. Second, this is the honest version of the video's compounding story. Something does compound — but it's the audience and credibility a landed game builds, not the act of shipping. A mailing list of zero compounds to zero.
Two cases make the ladder concrete. The video cites Sokpop, the Dutch collective with ~90 tiny games, as "dozens of releases and then a massive hit like Stacklands" — but their actual sequence shows their second game (Simmiland, 2018) was already a ~$300K hit; the volume strategy was funded by early traction, not a substitute for it. And a developer like Gagonfe — hit at release #10 with six mid-traction games before it — is real but rare: across the entire catalog only ten developers got their first hit at release #8 or later, and nearly all of them climbed through the mid rungs on the way. The many-quiet-games-then-jackpot career essentially does not exist. The many-mid-games-then-hit career does, and it's precisely what the ladder predicts.
After a miss: pivot beats doubling down
Everything above is about reading the market; the rest of this post is about decisions, because "stay in the game" is only half an instruction — the other half is what to change between games. Suppose your game missed: under 50 reviews, nothing better behind you. The data on what to do next is unambiguous. Developers who stayed in the same genre got their next game to a foothold (50+ reviews) 5.7% of the time; developers who pivoted to a different genre: 12.0%; a clean break sharing at most one tag with the miss: 13.5%. The single worst move measured anywhere in this analysis is a sequel to the miss: 3.2%, with zero hits in 218 attempts.
The toggle worth pressing here is LOVED: filter to misses that players adored (85%+ positive reviews) and… nothing changes. A beloved miss climbs at the same rate as a disliked one, and even a loved miss is better abandoned than doubled down on (10.0% vs 17.7%). Below ~50 reviews, positivity simply carries no information about your next game — the 40 people who loved it are real people, but they are not a market signal.
The same logic shows up in what breakout games look like. Taking every developer whose first-ever hit came from a mid-traction portfolio and comparing them to identical portfolios that didn't break out: the climbers priced up (59% raised their price at least 1.25× over their own norm, vs 28% of non-climbers; median breakout price $12.99 vs $4.99), changed genre (only 12% kept their usual primary tag, vs 37%), shipped new IP rather than sequels, and took longer than their own usual gap. The breakout game is almost never "the same thing again, but better."
Time off is free — at every level
One more claim you can stop worrying about: the gap between your games. The chart below shows next-game hit odds by how long a developer waited, at three career rungs, and it's monotone in favor of long gaps everywhere — after a hit, a follow-up within 6 months converts at 24% while one after 4+ years converts at 57%; even after a miss, the 4-year-gap developers do 14× better than the 6-month ones. Hover the bars and watch the median price climb with the gap: the gap isn't magic, it's a proxy for how serious the next game is. But the negative result is real and unconditional: there is no absence penalty anywhere in the data. The market does not forget you, at any rung. Nobody's window closes because they took three years off — the trap in the data is the opposite pattern, the rapid-fire shipper.
Volume vs craft, per year of your life
So should you ship many small games or one big one? Normalize careers by time rather than by game and the answer splits by what you're optimizing. Among serious developers (at least one game over 50 reviews, careers started 2015–2021): the best odds of ever landing a hit belong to craft-with-persistence — one real game every one to three years — at 40.7%. The most hits per year of career belongs to the 1–3-a-year middle. And the flood (3+ releases a year) buys the steadiest income floor (median $28.6K/yr) at the cost of the ceiling and half the hit probability. Each cadence wins a different game; pick which prize you're playing for. (Sokpop, for what it's worth, earns about 3× the 90th percentile of their own cadence class — cite them as an outlier, not a strategy.)
What keeps developers going (nothing you'd expect)
Given how much survival matters, we looked for what predicts shipping a second game at all. The strange, quietly profound answer: outcomes don't. Developers whose first game missed continue at 17%; mid traction, 23%; a full hit, 20% — nearly flat. Positivity doesn't predict it. Estimated revenue doesn't predict it, even at equal review counts and equal positivity. As far as external validation is measurable, persistence looks like a trait people arrive with, not a response to results — which is, in a sense, the video's "the ones who last love what they're doing" claim, measured from the outside.
What does move the needle is project design: cheap first games keep their developers alive (under $3: 19.4% continue; $12–20: 14.8%), big-scope genres burn them out (RPG/Simulation/Strategy starters continue least), and the one measured killer is launching your first game into Early Access — 8.7% continuation, half the baseline, because the unfinished game traps its developer inside it. If you want to still exist after game one: small, cheap, finishable, not Early Access.
Where newcomers actually get on the ladder
Since the first rung matters so much, here's where first games reach it. Each dot below is a genre (by primary tag, first games 2019–2024): horizontal = how often a newcomer's game reaches 50 reviews, vertical = how often a game that gets there becomes a hit. Up-and-right is where you want to start: Horror (34% on the ladder, 32% of those hit) and Simulation, with Visual Novel the volume-friendly surprise. Bottom-left is the graveyard: Casual, generic Action, and — worst in the entire catalog on both axes — Precision Platformer. The two gray dots are artifacts, not advice: Free to Play's "hits" are a reviews-per-download illusion, and Early Access marks seriousness rather than causing success.
One warning this chart can't print on itself: it's backward-looking by construction. A lane that opened recently won't show career value yet because its developers haven't had time to have careers — the current "short incremental" wave being the obvious example this data can't see. Genre tables tell you about the market that was; they're a floor, not a forecast.
What your audience is worth next time
Finally, the ladder's mechanism, made concrete. Collecting every case where a developer with an existing review base launched their next game: the median next launch opens with 4–7% of your total review base showing up as first-month reviews, and the chance of opening warm (50+ first-month reviews) climbs from 9.5% at a sub-200 base to 70% past 20K. Then flip to the LAUNCH => HIT view for why that matters: a game that gets 100–249 reviews in its first 30 days ends up a 25k-copy hit 54% of the time; past 250, 94%. The first month is nearly destiny — which is exactly why the compounding asset is the audience that shows up to it.
And a detail that surprised us: within equal audience sizes, longer gaps activate the audience better (58% warm-open after 3+ years away vs 19% for quick follow-ups — the seriousness proxy again), and this site is its own small proof: SNKRX opened at 117 first-month reviews off BYTEPATH's ~200-review base, a 60% activation, and went on to sell about 182,000 copies on Steam, plus a mobile port.
The honest summary
Where the video is right, it's importantly right: the market is a power law; most developers quit; the lifetime survival curve really does read ~60% → 20–30%; staying in the game is necessary, because you can't land a rare event you're not present for. Where it's wrong, it's wrong in the way that changes what you should do: the odds do not climb with games shipped — the coin-toss-by-game-five claim is what survivorship bias looks like when it's presented as motivation, and its own author explains survivorship bias earlier in the same video.
What the data actually supports is a sharper instruction than "stay in the game": stay in, and climb. Land any rung of real traction — even a few hundred reviews multiplies your future. When a game misses, change more, not less: pivot the genre, raise the ambition, never sequel a miss. Take the time the better game needs; nobody is forgotten. Design your first projects so you survive them — small, cheap, finished. And treat the audience a landed game builds as the one asset that compounds, because it's the only thing in 48,049 careers that measurably does.
Every number here comes from the primary data — Steam's own endpoints, ~35K review-history crawls, and a revenue estimator calibrated on this site's own games. The full research corpus behind this post (31 reports of it) was built across several sessions with Claude, whose complete logs are on this site: 1, 2, 3, and 4 — the session that checked the video and built this post, charts and all. If you want a claim checked against the data that isn't here, say so in the site chat.