Unpredictable Hits #011: UA Became Lab Work + 1,000 Ads For A Single Launch
From the Editor
What's up friend?
Two numbers from this week belong side by side. Monthly game launches doubled over the past year, but the number of games that actually broke through didn't budge. Twice as many swings, and the same short list of hits.
That gap is the whole story right now. Making a game got cheap, and then making the ads that sell it got cheaper too. Once everyone can produce more for less, producing more stops being an edge.
Grand Games launched a single game with over a thousand creatives and 260 playables. When the studio across the street can flood the same auction you're bidding into, volume stops being the advantage, and the edge belongs to whoever reads the results fastest and acts on them.
That's quietly what the job has become. The networks took over bidding, targeting, and budget pacing, and what they left in your hands looks a lot like lab work. You form a hypothesis, test it at volume, read the signal, kill what's dead, and run it again. The main piece this issue is built around exactly that idea: the UA manager as a scientist rather than a buyer.
Also inside, you'll find the recording from my webinar with Luis de la Cรกmara and Gil Tov-Li, and nine things from the week that all point in the same direction. One of them is a Naavik report that made my month.
Featured: UA as a Scientist
For about a decade, creative ideation was the part of the job where someone had a good idea and you bet on it. Taste, instinct, a creative director's gut. That version of the work is gone, because the ad networks stopped being tools you operate and became recommendation engines that decide who sees what. Once the machine owns targeting, the only input still fully in your hands is the creative, which means the creative is now the targeting.
You can't reason your way to the right creative, any more than a chemist can think their way to the right compound. You have to search for it. Every concept is a hypothesis, most of them are wrong, and the skill that's quietly becoming more valuable is knowing how to branch toward the few that work and cut the dead ends without mourning them. There's already a word for working this way. It's science.
The full piece walks through what that job actually looks like: the tree of hypotheses you're really searching, the six places good ideas come from, and why a single winning ad was enough for one studio to build a whole new game around it.
Webinar Recap: Game Growth is Changing - Featuring Luis De La Camara (Rovio) & Gil Tov-Li (AppCharge)
Playtika just told its investors that roughly 40% of its revenue comes from direct-to-consumer. That number isn't an outlier anymore. It's becoming the floor, and it's the stat that opened our latest webinar. We put two people on the same call who see the shift from opposite ends: Gil Tov-Li, CMO of AppCharge, a D2C platform processing a billion dollars a year for studios like King and TripleDot, and Luis de la Camara, VP of Global Marketing at Rovio.
The web store stopped being a side project. A purchase inside the app gives 30% to Apple or Google. The same purchase through a web store costs four or five percent. "The cat is out of the bag," Gil said, "and there's no coming back." The studios doing it well now run close to fifty-fifty: half the revenue through in-game payment links, half through the web store. The payment link is the on-ramp that teaches a player the web store exists.
Then we turned to the part that keeps every UA lead up at night. Since AppLovin rebuilt its engine in 2022, the networks are starving for creative, and the only way to crack them is enormous volume in enormous variety. Look at the top of the charts and you'll find APAC studios testing fifty, sixty, seventy playable concepts a month plus thousands of videos. Luis had the sharpest line of the day: the edge is speed of production. "AI gives a creative person superpowers," but "the human in the middle still outperforms the fully automated version." Picture two hundred in-house creatives, each one supercharged by AI. That's the wall the West is bidding against, and it's the exact problem we're obsessed with at Sett.
Gil's hot take landed hardest. The industry pours its best UA talent into one problem, acquiring and re-acquiring users, and almost none of it into getting the existing base to buy more. "You're not going to beat the two-hundred-person teams at their own game," he said. "Find better arenas to do battle in." The population is flat. The studios that win will be the ones that monetize the audience they already have.
We closed by asking each of them for one focus for 2026. Luis refused to frame it as a retreat from UA: "In my world, it's not an OR. It's an AND." UA stays as the channel that captures players ready to play now, and the question is what you build around it, from D2C margin to brand to the Angry Birds movie in theaters this Christmas. Gil pointed at the tooling: get literate with AI agents and Claude Code, your whole team, not just you. Twenty years in, the constant is the hustle, and the winners are the ones who move first.
Industry Buzz
1. Naavik Maps the AI-Creative Stack. The Moat is Orchestration.
Author: Francois Courset, Naavik
Naavik set out to map the state of AI for mobile ad creative, and Sett shows up across the piece. The line worth your attention isn't about funding, it's about where the moat sits: "The technological moat is typically concentrated in the agentic orchestration layer (Sett) as asset generation models tend to become commoditized." The models that crank out a single asset are becoming a commodity. The durable edge is in the agents that orchestrate the whole system. Naavik also notes Sett "is already working on ad deployment agents," and points to a client result with Supersonic across output quantity, quality, and share of voice.
Underneath the write-up is the math that explains why this matters. Top spenders now ship 2,400 to 2,600 creative variations a quarter, up 25% to 30% year over year, and 56 of the top 100 grossing games used AI creatives in 2025. The number I keep coming back to is the gap in how people feel about it: 82% of ad execs are comfortable with AI ads, and only 45% of players are. That gap is the real story in the report.
2. Volume Got Cheap. Winning With It Didn't.
Author: Aviad Dadon, Gamigion
Gamigion's 2026 outlook lists seven shifts, then admits they're really one thing wearing seven outfits: "These aren't seven separate trends. They're one system rearranging itself."
The shift that matters to us is the simplest one in the set. Once AI made creative free, volume stopped being the thing that sets anyone apart: "Creative got cheaper to produce before it got easier to win with. When everyone can generate volume, differentiation becomes the moat." Then comes a line that's basically our thesis in someone else's mouth: "What matters now is not whether a studio can produce 50 creatives. It's whether it understands what the best ones are actually saying about the player."
The rest of the shifts rhyme with that one. Hybrid monetization is the default now, DTC has graduated from a side experiment into a core revenue layer (Stillfront ran 44% of Q1 bookings through it at 84% gross margin), retention-led growth is eating install-volume growth as mobile downloads fell 7% in 2025, and the studios rebuilding around AI instead of bolting it on are the ones compounding.
3. Grand Games Went From 20 Creatives to 1,000 for One Launch.
Author: Matej Lancaric
Here's the arms race up close. Grand Games used to launch with around 20 creatives. For Block Out, they shipped more than a thousand creatives and 260 playables, which is roughly fifty times their old output for a single release.
And it's working. They're pulling in roughly $300K a day and 140K iOS downloads a day, with $70M raised last month and over $105M in total. Lancaric's read on why they keep winning is worth pinning up: "They don't do giant innovation. They do perfect execution." Block Out is a cleaner, sharper Color Block Jam with better level design, and they happily let originality take a back seat to volume and craft.
One line in the piece belongs above your desk: "A playable is not an ad, it's a tiny game." If you're still treating playables like banners, you're losing to the teams who treat them like products.
4. Launches Doubled. The Winners Didn't.
Author: Matthew Emery, Gamigion
This is the companion to the Grand Games story, and it's where the stat at the top of this email comes from. Monthly mobile game launches doubled year over year: "Starting in Sept last year, monthly mobile game releases began to climb to unprecedented numbers."
All those extra launches didn't buy a single extra winner. "This, unfortunately, does not mean that more games are succeeding." The slice of games clearing even a 10k-download bar actually got thinner. Twice the launches, the same short winners' list, and the gap between shipping a game and winning with one is exactly what creative volume and experiment discipline are fighting over. Item 3 is the answer to who's winning it.
๐ Read Matthew's piece on Gamigion
5. Match Villains Built a Game to Fit the Ad. Reverse UA is Here.
Author: Matej Lancaric, Gamigion
Normally you build the game first and then test creatives against it. Good Job Games ran the whole sequence backwards. A police-escape ad for Match Villains tested too well to waste on a match-3 store page, so they built a separate game, Police Escape! Match Villains, to deliver what the ad had already promised.
"The ad came first. The game came second." And here's the part that should make every UA lead sit up straight: "They didn't guess whether police chase converts. They built a game to fit a creative they already knew would win."
This is the scientist mindset with the safety off. The creative becomes the experiment, and one winning result is now enough to greenlight an entire new game.
6. FunPlus, ~2,000 People, Going "AI-Native".
Author: Miriam Ahmad, PocketGamer.biz
FunPlus, a publisher of close to 2,000 people, is rewiring the entire company around AI, baking it into finance, HR, and engineering as basic infrastructure rather than a marketing toy. CBO Chris Petrovic states the ambition plainly, "Our goal is to be an AI-native game company," and he sets the bar for everyone else just as plainly: "If you're not doing it, your competitors will be."
The tell that it's real is a small one. Their internal AI-sharing channel was "almost impossible to keep up with" in the first year, and then it went quiet, not because people lost interest but because the tech stopped being a novelty worth posting about and turned into how the work simply gets done. That's the moment any technology actually matters, when it stops being news and becomes plumbing.
7. The 2026 Retention Floor: 47 / 24 / 13.
Author: Matthew Emery (LinkedIn)
The cleanest answer going around to "what counts as good retention right now" is this: for a top-grossing-tier match-3 in 2026 on US iOS, the install-weighted average lands at D1 47, D7 24, D30 13.
There's a small surprise sitting underneath it. Some casual genres still work on a D30 of 5% to 8%, which I genuinely did not see coming. Solitaire and Tripeaks run modest profiles, and Disney Solitaire is "a SERIOUS outlier" that ate its genre.
Retention is the floor your creative pours users onto, which is why it can't live on a separate scoreboard. You can flood the top with a thousand creatives, but if the game leaks on day two you're just filling a bucket with a hole in it. Volume and retention are one system, and the teams that treat them as two lose on both.
๐ Check Matthew's post on LinkedIn
8. Asia's $103B Market is Adding Players Faster Than Revenue.
Author: Aylin Yazฤฑcฤฑ, Deconstructor of Fun (on Niko Partners' 2026 Market Model)
Here's a useful gut check on where the growth actually lives. The 13 markets in Niko's model did $88.9B in 2025 and are on track for $103.6B by 2030, a 3.1% CAGR that works out to roughly 46.3% of global game revenue.
The line to read twice: "The growth you keep hearing about in these markets is player growth, and that is a very different thing from revenue growth." India is growing 11.2%, but at around $2 per player against $371 per player in Korea, while China, Japan, and Korea still hold 88.6% of the region's revenue in 2030.
For UA, that quietly detonates the regional LTV model. You either set targets market by market, or you torch budget chasing installs that will never pay you back. Two more worth tracking on the side: women are now 42% of players, and mini games make up around 20% of mobile spend in China.
๐ Read on Deconstructor of Fun
9. The May Scoreboard - Honor of Kings Runs Away, Arrow puzzles Storm Downloads
Author: Dmitriy Byshonkov, GameDevReports (data: AppMagic)
The May charts finally moved. "Finally, interesting movement in both Downloads and Revenue charts."
The top five by revenue: Honor of Kings at $155.9M and miles out front, Whiteout Survival at $105.2M, Royal Match at $105.1M, Gossip Harbor at $96.8M on its second-best month ever, and Last War: Survival at $80.7M, with Brawl Stars climbing into the monthly top 15 at $38.2M a little further down.
The faller is the more interesting story. Last War's IAP dropped 30% in May to that $80.7M, its lowest monthly take since March 2024, and part of the reason is that it's steering players to DTC, which sits outside store-tracked revenue, so the public numbers keep undercounting the real haul. Over on the downloads side, arrow-style puzzle games swept the top, with Arrows: Puzzle Escape at 28.6M and Arrows GO! at 23.8M. A genre coming out of nowhere on the back of fresh creative is the "explore smarter" thesis printed straight onto a chart. (Revenue here leaves out Android from Chinese stores.)
Wrap Up
Put the whole week in one room and it tells a single story. Launches doubled while the winners' list refused to grow. Grand Games answered with a thousand creatives and 260 playables for one game. Match Villains built an entire product to match one winning ad. Naavik drew the category map and set Sett's orchestration layer in the middle of it. FunPlus is rewiring 2,000 people around AI. Different starting points, but they all land in the same place: volume is free now, so the edge is the experiment.
That's the job. You're running a lab. The networks kept every lever they could automate and handed you the one that decides who actually wins.
We get into all of it in the UA-as-a-scientist piece, and Luis, Gil, and I went deep on the growth side in the webinar. Both are up above.
See you next time.
Yours, Jonathan (Fishi) Fishman
Head of Marketing, Sett
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 youโre still counting sheep.
