Unpredictable Hits #014: 100 of 220 Employees Making Playable Ads + 81% of UA Spend Flows Through Playables
Hey everyone, greetings from Beijing!
Amit, Sett's CEO, and I are traveling throughout China for China Joy and meeting some of the world's best mobile game studios based here. It's our first time in China and it definitely gets one thinking.
The game industry in China that markets to the West (overseas marketing in their terms) is one of the most sophisticated and advanced ones in the world, with hundreds of game studios operating at an unmatched pace and level of focus. It's not surprising that these studios are consistently topping the charts.
One thing that's apparent is how user acquisition and creatives are one of the central pieces in these studios - treating them at the same perfection level (if not more) of care and attention to details that the game itself is receiving.
One studio we met has 220 employees, sitting in a beautiful Beijing office, and 100 of them are making playable ads. When their game was launched, it took them three weeks to reach a six figure UA spend level on a daily basis.
I will write a more detailed note on what Western studios can learn from the Chinese ones when we're back in Tel Aviv, and also how AI adoption is going in these studios - so stay tuned.
In the meantime, lots to cover this time, so let's dive in.
- We published The State of Playables, our data report on how winning studios use playables. The short version is that 81% of the market's UA spend flows through the games that leverage playables.
- We sat down with Peerplay's Yotam Pappo and heard why he thinks the MVP is dead.
- Singular 2026 AI growth guide, with Sett's own contribution.
- Toon Blast's downloads are down more than 90% from peak and it's the best business in puzzle.
- Sensor Tower put out its first ever ad monetization report and landed on $12 billion, a number the analyst who built it calls a floor.
- Naavik found a 2% revenue dip hiding a 12% download drop. July's US top ten downloads went seven-for-ten to Chinese and Turkish studios. And Aylin Yazฤฑcฤฑ explains why the same publishers keep winning genres that look wide open.
LFG.
Sett Stuff and Analyses
The State of Playables: 20% of the Games. 81% of the Money.
We pulled the data on 16,296 mobile games from the world's top studios and $6.28 billion in trailing-3-month UA spend to answer one question: where do playables really sit in the market right now?
The headline: Only 20% of the games run playables, but 81% of all UA spend flows through the games that do. The debate about whether playables will work for your game is over. The market shows the truth.
The ladder: Adoption climbs with budget, from 16% of the games spending under $100K a quarter, to 59% between $100K and $1M, to 81% between $1M and $10M, and 84% above that. Among the roughly 9,800 games running minimal UA, just 2% touch playables. The pattern holds across nearly every genre once budget is equal, with RPG the one outlier.
The volume story: Among adopters the median library is 32 playables, but 130 games run a thousand or more each, and the top adopters average 332 while still making close to three times more videos. Nobody at the top treats a playable as a one-off.
The acceleration: A year ago 1 in 8 new ad creatives in mobile gaming was a playable; by May it was 1 in 3. At that pace the market ships a new playable roughly every 35 seconds.
The message: If you don't have a way to create new playable concepts at volume, you're capping your game's UA growth at your own production capacity. The studios at the top of the charts are out-exploring you.
Oh, and by the way, if you want your own genre's numbers, there's a one-pager inside.
๐ Read The State of Playables
Yotam Pappo on Pods of One and the End of the MVP
In season three, episode two of our podcast, Unpredictable Hits, we sat down with Yotam Pappo, co-founder and CEO of Peerplay, the AI-native casual studio he founded in early 2023. Its first game is Merge Cruise. A few things worth taking from it:
Pods of one: AI collapsed the size of a team unit. In Yotam's words, "you have pods of one", one person owning a slice end to end. It only works if senior people accept being beginners again: "You can't have a culture of curiosity and discovery without people feeling that they're juniors."
Genres are our map, not theirs: "I don't think that users play genres." An interesting take, worth reading against items 4 and 5 below, where the market gets sliced by genre all the way down.
The MVP is dead: When building gets this cheap, a deliberately thin first version stops being the smart opening move. Yotam argues for betting on vision over letting early data make the call.
๐ Watch or listen to the full episode
Singular's AI Growth Guide: Everyone Bought the Tools, Few Changed How They Decide
Singular published its Essential Growth Guide for 2026, "Marketer's Guide to Turning AI into Real Growth." It's a 36-page look at how high-performing marketing teams use AI right now, full disclosure: one of the talking heads in there is me.
Where the market really is: 49% of marketing teams already use AI in their workflow, but only 8% point it at analyzing data, and 66% report minimal ROI so far.
The Sett piece: Our spread is titled "Build better exploration systems, not better crystal balls." The argument: most teams adopt AI hoping it will tell them what's going to work. In mobile gaming UA, 97% of creative ideas fail, and AI doesn't change that rate. What it changes is the speed at which you move through the ones that don't work to find the ones that do. In the guide's words: "The old model was opinion-driven. A creative director picks three concepts, the team spends two weeks producing them, and maybe one kind of works." Leading studios already explore 50 to 100 genuinely different playable ad concepts per month.
My line from the guide: "AI makes uncertainty cheaper to navigate. The cost of being wrong on a single creative drops dramatically when you can produce and test 10x the concepts in the same time window."
Industry stuff worth knowing
1. Toon Blast Is the Best Business in Puzzle. Downloads Are Down More Than 90%.
Author: Aylin Yazฤฑcฤฑ, Deconstructor of Fun
A nine-year-old game with downloads down more than 90% from peak is pulling revenue near record highs. Toon Blast went from $135M in H1 2023, visibly on its way out, to $212.7M in H1 2024, and has held above $35M a month since, more than double its 2023 average. US ARPDAU climbed from the 17-38 cent range to $2-3, which works out to the same install base at roughly ten times the value per user.
The trigger: Peak adapted Royal Match's Super Light Ball feature in about three months, while Candy Crush still runs a heavily diluted version. In Yazฤฑcฤฑ's words: "it isn't a generosity feature, it's difficulty headroom wearing a costume."
The bigger lever: The Puzzly Monthly panel behind the piece splits the credit roughly one third to Super Light Ball, two thirds to everything else, with the biggest piece of the everything being a rebuilt events calendar: seven isolated one-day events replaced with overlapping two-to-six-day formats.
The UA kicker: With per-user economics up 10x, the article argues someone inside should be making the case for 10x the UA spend. And the closing shot lands on everyone: "Most studios in this market already have a Toon Blast sitting in their portfolio. They're just calling it a legacy title and quietly moving their best people off it."
๐ Read on Deconstructor of Fun
2. Sensor Tower's First Ad Monetization Report: Every Number Is a Floor
Author: Josh Chandley, President & COO of WildCard Games, on Deconstructor of Fun (from his interview with Sensor Tower's Lead Analyst, Sam Aune)
Sensor Tower published the first ad monetization estimates in its history: $12 billion in 2025 mobile game ad revenue, from 2.4 trillion impressions, across 19 countries. Chandley's point is that the number is a floor, and the analyst who built the report agrees. "The floor analogy works best," Aune told him. Chandley's own estimate is $16-20 billion, and he's spent a decade buying and selling this inventory, so take the point.
The share: About 23% of net mobile game revenue in the measured countries comes from ads. In the US, with web-store D2C counted, the split runs 73% IAP, 14% web store, 13% ads. Side income stopped being the right word a while ago.
The new money: Non-gaming advertisers, Temu, Rocket Money, ChatGPT and friends, now take 30-50% of impressions in every game genre while paying just 28% of the revenue, mostly high-volume banner demand filling placements gaming advertisers were never going to buy. Chandley remembers Walmart as WildCard's number one advertiser by impressions.
Nine figures on ads alone: Block Blast has done about $127M (although I strongly suspect this number is much bigger) and Vita Mahjong about $90M so far in 2026, per Sensor Tower, with no meaningful purchase economy behind either. The unsolved problem, per Chandley, is an IAA-first hybrid that doesn't cannibalize itself, because "In a game that monetizes attention, the store's best product is already free."
๐ Read on Deconstructor of Fun
3. Naavik: Mobile Gaming's Stability Is an Illusion
Authors: Harshal Karvande (Principal Designer at Rovio) and Abhimanyu Kumar (Co-founder of Naavik)
H1 2026 revenue came in around $28 billion (Naavik analysis, excluding D2C and China Android and more - it's a subset of the market that Sensor Tower puts at roughly $40B IAP in H1, and it's most likely bigger than that, but Naavik is analyzing a subset to show the representative dynamics), down just 2% year over year, which looks stable right up until you get to downloads: down 12%, with Q2 alone back at what the authors call pre-pandemic levels. An 11% rise in revenue per download papered over the gap.
The aging problem: Games five years and older now generate 57% of revenue among the top 200 grossing titles, and around 70% of total mobile gaming revenue. New games under two years old replaced 99% of what the incumbents lost this half, but only about 1 in 4 young top-200 entrants ever becomes a long-term incumbent.
The UA squeeze: Gaming digital ad spend rose 8% year over year while impressions rose 14%. The authors read the spending climb against shrinking installs as CPIs likely rising faster than LTV can improve. Every UA leader reading this already knew, but now there's a chart.
The verdict: The authors land close to Supercell CEO Ilkka Paananen's line from his annual letter: "It's an industry coastingโฆgetting very good at optimizing what already exists."
The genre split: Puzzle was the standout this half, up $915M year over year, while RPG, Casino, Strategy, and Action bled a combined $1.6B.
4. July's US Download Chart Belongs to China and Turkey
Author: Gรถkhan รzmez, Gamigion
Three Turkish games cracked the top four of July's US iOS download chart: Smash Fest! by Flow Games (2.67M+ downloads), plus Block Out! (2.09M+) and Magic Sort! (1.69M+), both from Grand. The #1 spot belongs to Meowdoku by Learnings, last issue's Minesweeper-meets-Sudoku cat machine, at 3.3M+.
The composition: Per Gamigion's list, 7 of the top 10 most-downloaded games in the US in July come from Chinese or Turkish studios, four Chinese (Meowdoku, Vita Mahjong, Block Blast!, Gossip Harbor) and three Turkish. In รzmez's words: "Chinese presence is significantly high."
His read: Everyone calls these games hybrid casual because of the ad revenue. รzmez thinks they're just puzzles, and points at the low-CPI opening they exploit. His homework for you: go check how many CPI videos and playable ads Meowdoku ships in a month.
Where I'm sitting: I'm writing this issue from the country holding four of those ten spots.
5. Why the Same Studios Keep Winning
Author: Aylin Yazฤฑcฤฑ again, this time in her own Games & Takes newsletter. She's having a good week.
Yazฤฑcฤฑ went through AppMagic's H1 2026 report and found the casual market flat at around $11 billion while nearly every big publisher in it keeps growing. The two facts only reconcile one way.
Match-3 slammed shut: 120 new Match-3 games launched this year and not one of them passed $100K in monthly IAP revenue even once. The genre did $2.5B in H1, flat, with downloads up 10%, and the winners keep winning: Royal Kingdom has out-earned Royal Match over the equivalent post-launch window.
Merge looks open. It's not: The genre grew 74% year over year to $1.3B. Seven new games passed the $100K monthly bar, and five of them came from Century Games and Moon Active, publishers already successful in the genre. Century's Tasty Travels did $127M in H1 alone, up more than 13x.
Her thesis, in her words: "you're no longer competing against a game, you're competing against an organization." The moat is years of accumulated LiveOps, economy, CRM, and UA expertise. Her closing:
"A great idea might still get you noticed.
A great organization is what keeps you relevant."
Wrap Up
I keep coming back to the studios we visited in China, where a hundred of the two hundred and twenty people on payroll spend their days making playable ads. That ratio would be unfathomable in most Western studios. There it was simply how the company was built.
The Western industry keeps asking what tools the winning studios use, as if the answer is a subscription you can buy. What I saw this week says the real decision happens on the org chart. The studios eating the US charts staffed creative exploration like a department, with headcount and budget, and treating it like high quality manufacturing, while in most Western studios it remains a side quest.
It's not easy to achieve, but the teams doing it anyway are finding winners faster, and the charts don't care how anyone feels about it.
That's what Sett is for - an exploration department delivered as AI agents that generate playable concepts at scale, read the results, and keep hunting until they find the fun your UA can scale on - and you don't need a hundred-person floor in Beijing to get it.
Your competition already made this decision. All that's left is how long your org chart takes to catch up.
See you next time.
Yours, Jonathan (ๆธธๆตท) 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 youโre still counting sheep.
