Unpredictable Hits #019: How Block Blast Decides When Players Lose + 158 Games Past $1M
What's up friend?
You probably saw this one. Apple's review queue, the approval a new game needs before it goes on the App Store, went from the usual 24 to 48 hours to weeks in some cases, according to the studios who spoke to MobileGamer.biz (Google Play didn't budge, by the way). The suspected cause is a flood of AI-made apps. Apple's side of it: 378,000 submissions a week, a 1.5-day average, and updates to live games go first.
That's the theme this time: making a game has never been cheaper, so there are more of them than ever, and the scarce thing is knowing which one deserves real money, and how long to wait before you decide. A prompt now makes a playable game in twenty minutes, and almost nobody plays it twice. 158 games reached $1M this year, and most of the originals needed three months to get there. And our Barcelona panel is the same problem one level down, at the creative: what has to be in place before a winner can show up at all.
This edition:
• UATOPIA session - a recorded panel from our UA event in Barcelona two weeks ago: ex-Rovio and King Luis de la Camara and Playtika's Guy Bauer on what has to be true before a winner shows up. First session out, for free.
• A new Unpredictable Hits podcast: the Mad Brain Games founders keep 40 games in production with zero artists and launch one a week.
• 158 games crossed $1M in 2026, and only 7 of the 116 originals did it inside a month.
• A game from a prompt in 20 minutes, and 3% of players still there a month later.
• What good looks like across the top 50 puzzle games.
• Block Blast puts a price on each of its 48 million daily players, then decides when you lose.
Yalla.
🔥 Sett stuff and analyses
UATOPIA panel: What has to be true before a winner shows up
Two weeks ago in Barcelona, Luis de la Camara (ex-Rovio, Scopely, King) and Playtika's Guy Bauer sat down with Sett's CEO Amit Carmi for an hour on the question every UA team argues about: when does a test win deserve real money? They touched on feedback loops and how to get more from the creative team, brand versus gameplay in the creative, the fake-ads argument, and the creatives who understand UA and the ones you'd let go.
Guy set his table with a loop: small tests that feed straight back to the creative team, because "I don't need to spend a lot to get the indication," and micro teams of experts over volume, because "I never believed in quantity." Luis set his with the fights that come before any test: a testing budget carved out with the CFO that is allowed to not return, boundaries you only find by crossing them (an engineer once threatened to quit over "fake ads"), and UA people who keep their ego out of the performance so the learning compounds. One rule he won't negotiate: you can't be good at UA creative without understanding how the networks score it.
This is the first of the recorded sessions to go public. I'll share more in the next newsletter.
New on the podcast: 40 games in production, zero artists, a launch every week
Tomer Waintroub and Avi Asher met at Playtika, have started every day with the same 9:00 a.m. call for 14 years, and now run Mad Brain Games on a platform that keeps more than 40 games in production with no artists on staff and a new launch every week.
Amit asked them where the ideas come from (the ads they play, before the store charts), why they built their own incentivized channel and the fraud fight that came with it, and which jobs they handed to AI agents with product, marketing, monetization and R&D titles, along with the exact point where they stop trusting them.
🔥 Industry stuff you should know to be better
1. 158 games crossed $1M this year, and the slow ones are winning
Author: Andrei Zubov, AppMagic, on Gamigion
AppMagic took every game released between January 1 and October 13 that passed $1M in lifetime IAP revenue by October 17, and got 158 titles once the China-majority games were removed.
The numbers that matter:
- IP-based games: 42 of them, and 20 reached $1M inside 30 days. Original games: 116, and 7 did.
- Puzzle took the longest, a median of 126 days and none inside a month, and ended with the highest median lifetime revenue at $4.3M. Merge-2 and Sort are 17 of those 26 puzzle titles and 80% of their revenue.
- North America-led games needed a median of 89 days, and only 2 of 60 made it in a month. Those two have earned $455.3M between them, more than half the whole sample.
- Jelly Busters took 144 days to reach $1M and did $5.8M in the last 30 days.
The line to keep: "a slow first million can still be followed by much stronger growth." The sample only includes games that already passed $1M, so it can't tell you how many never got there. It can tell you that a 30-day verdict would have killed most of this year's puzzle winners.
2. A game from a prompt in 20 minutes, and nobody plays it twice
Author: Matthew Camp, Naavik
Naavik's Matthew Camp tried five prompt-to-game tools: Roblox Build, Meta's Horizon Create, Google's Playground, and two startups, Astrocade and Sekai. You type what you want, and each one hands back a playable game in under 20 minutes.
- Roblox scored one of its own Build games 100 out of 100 on the automated playtest and 31 out of 100 with design judges, with zero for replayability. Roblox's own words: it "isn't a game that anyone would choose to play twice."
- Astrocade and Sekai installers show about 3% day-30 retention on Android, against about 17% for Roblox, and Roblox users spent 13 to 14 times longer in the app.
- 58% of Astrocade's Android downloads from April to September came from paid ads. Its CEO says 10M monthly users. Sensor Tower counts about 0.8M on Android.
Naavik's read: more games doesn't mean more play, and discovery still concentrates on hits. Roblox is best placed because it already has the audience, and the startups have to prove prompt-made games can hold players at all. That's the flood from the top of this email, with numbers on it.
3. What good looks like in puzzle, from the top 50
Author: Anton Slashcev, on Gamigion, with AppMagic data
Slashcev pulled retention, playtime and monetization for the top 50 puzzle games. The findings:
- Median D1 retention 42.8%, D7 25.0%, D30 15.7%.
- 77 minutes of playtime per active user per day.
- $4.80 lifetime revenue per download, $0.26 ARPDAU.
The outliers are the fun part. Candy Crush Soda has the weakest D1 in the set at 26.5% and still made it. Farm Heroes Saga leads both D7 and D30, years after launch. LINE PokoPoko earns about 60 times the median ARPDAU. Bus Traffic Fever is last on both money metrics and is still in the top 50. His conclusion, and I'd agree: strong retention, deep monetization or pure scale can each get you there on its own. He doesn't say which months or countries the data covers, so use these numbers to compare, and don't chase them.
4. Block Blast decides when you lose
Source: Ayo Omolewa's teardown, with Felix Braberg's read
Block Blast is the free block puzzle that about 48 million people play every day (Sensor Tower's count, via Felix), and it lives on ads, with a rewarded ad as the way back when you're stuck. Ayo Omolewa, an independent data scientist, pulled its logic apart. His first finding is the one that matters most: most of the rules aren't in the app you download. The phone fetches them from the game's server after you install, so the build Apple reviews doesn't contain them.
The rules, in the order they reach you:
- It prices you. A model with 27 inputs estimates what an ad shown to you is worth, mostly from the ads it has already loaded, plus your country, your phone and how your first game went. Ayo ran a new iPhone player with a first score of 1,000 through it: about $25 per thousand ad views in the US, about $8 in Brazil. One oddity Felix pulled out: on iPhone, a weaker first game predicted a higher ad value, his guess being that it signals a casual, older player advertisers pay more to reach.
- It picks the moment. In Felix's words, "almost every rule that ends a game kicks in at the same moment: right after you beat your high score." Once you're 10% past your record, every new set of pieces has a one-in-five chance of being a set that fits nowhere.
- It waits for the expensive ad. A second rule fires only when three things are true at once: you finished your last three rewarded ads, you're about 1,000 points past your record (the bar varies by test group), and the ad it has loaded is worth at least 3.5 times the last one you watched. Then it deals the dead set, you lose, and "watch an ad to continue" is the way back.
- It knows when to go easy. Two games below your record and the pieces get friendlier. A separate model predicts whether you'll play less tomorrow, and if it thinks so, your next revive is free, no ad, once a day.
The money: Sensor Tower puts Block Blast's ad revenue at an estimated $127M for January to May, the figure Ayo cites. Felix works it out to $700K to $1M a day, which by his estimate makes it the world's largest single-app mobile publisher. Both numbers are estimates built on Sensor Tower data.
For UA people this is a CPI story as much as a monetization one. What a player earns inside the game sets what you can pay to buy them, and this game sets that number per player, in real time, before the second game is over. Felix's other warning is for everyone who publishes: they encrypted everything, kept it off the App Store build, and one person still pulled it apart and posted the code.
👉 Read Ayo's teardown, rule by rule
👉 Read Felix's take, with the revenue math
Quick hits
- Supercell completed its acquisition of Metacore, the Merge Mansion studio, on October 19. Terms undisclosed, 159 Helsinki staff cut on the way in, and AppMagic puts Merge Mansion at about $490M lifetime. Paananen says the game's "best days are still ahead." MobileGamer.biz
- MobileGamer.biz also rounded up the games in soft launch at the biggest publishers: Playrix has five in test, Zynga and its studios seven, Miniclip four, and Moon Active's Dream Bay is live in 11 European markets on Google Play. Worth a scan for what's about to hit the stores. MobileGamer.biz
- Matej Lancaric's tip of the week: "everything in UA is creative and/or geos," and if you still review creative on blended numbers once a week, "you're not doing creative strategy." He runs daily reports by channel and geo. Brutally Honest
Wrap up
One thing to try this week, in Ayo's spirit: play your own game for twenty minutes as a new player, on a phone that isn't yours, and write down every moment where the game made a decision for you. The hard level, the offer, the ad. Then ask who in the building can explain the rule behind each one. If the answer is nobody, that's a meeting to book immediately.
Till next time,
Jonathan (still in Apple's queue) Fishman
About the Author
Fishi is the Head of Marketing at Sett. His brain is a chaotic jukebox of ideas with more cultural references than any feed can handle. He collects sneakers and plays chess while you’re still counting sheep.
