Unpredictable Hits #015: 172 Core Games, Zero Western Winners + the UA Job That Just Died
What's up friend?
A career ladder just died in mobile gaming. The call came from the kind of person who does the employing. Amos Adler runs a growth org of around 50 at SciPlay, and he said it on our podcast with zero drama: "We don't need operators anymore. We need analysts. We need people who think."
The UA operator was the job title growth teams were built on for a decade. The networks took the dials years ago. The agents are now taking the routine, and what's left of the role is the part that was always hardest to hire for: knowing which question to ask next.
This edition:
- September 23, live: Madbox's Creative Director and me on what you do when your best playable stops working. Save your seat below.
- The full Amos episode, 47 minutes on the only lever in UA you fully control.
- Singular published its Creative Benchmark Report for 2026, and I pulled out the numbers worth arguing about.
- China's studios stopped debating AI and started shipping with it. The revenue share numbers are the argument.
- 172 core games launched in Q1, and every winner came out of East Asia.
- A Rovio creative producer wrote the field guide to adapting competitor ads without cloning them.
- Smash Fest set off 38 lookalikes in two months. The mechanic they're all copying is from 2018.
- One more pass at Sensor Tower's ad monetization report, this time the splits under the headline.
- And a pair of 95% stats that should not be able to coexist, from Aylin Yazฤฑcฤฑ's profile of Scenario's founder.
Yalla.
Sett Stuff and Analyses
Live on September 23: Your Best Playable Is Dying. Now what?
Every UA team knows the curve - your winner stops winning. You found a rare playable and ran variations until the lift flattened. Now there are three doors: squeeze the concept, re-theme it, or kill it and hunt something new. One budget to spend across them.
On September 23 I'm going live with Anton Neznamov, Creative Director at Madbox, to argue about what you actually do next. He brings the studio side. The dying winner is his, and so is the call on what replaces it. I bring the production side, where I watch playables die across dozens of studios and see what takes their place.
What we'll get into: telling a played-out concept apart from a worn-down audience or a network change. They look identical on a dashboard. Each needs a different fix. Plus how much of your output should keep squeezing the proven winner versus exploring the next one, and how a new concept brief stops being a coin flip.
The format: a moderated conversation with no deck and your questions at the end. It runs September 23 at 2 PM UTC / 4 PM CEST / 10 AM ET, it's free, and the recording goes to every registrant, so sign up even if the hour is impossible.
The Podcast: Amos Adler on the Only Lever You Control
Amit sits down with Amos Adler, VP of Growth at SciPlay and a graduate of the DauUp Mafia, the Israeli mobile marketing school that produced a serious chunk of the people running UA today. His market is social casino. Fresh users are scarce, recycling is constant, and the competitive set is sharp, which is exactly what makes his system worth hearing.
The lever: "With AppLovin, with Unity, with Facebook, with Google, creative is probably the only lever that you control 100%. Publishers, models, algorithms, it's always like a retroactive analysis compared to a proactive approach."
Money on the floor: SciPlay centralized creative strategy instead of letting 15 UA managers each run their own version of it. His summary: "We understood how much we're under-analyzing it. There's a lot of money on the floor there."
The attribution hunt: He treats attribution as a permanent hunt for the truth, not a setup decision. The hunt pays. "We slashed retargeting spend by 30% or so, and we lost less than 2% of revenue, which means how much more efficient you can be."
The forecast: "But commerce, CPA, IAA, I think that it will be 90% autonomous in the next couple of years, just because it's super predictable, and once something is very predictable, you already have loops to control content."
His advice for the people in those seats: "there's a wave coming, it can drown you, or you can ride on it, and that is being defined by how exploratory you are with AI."
๐ Listen to the full episode
Industry Stuff you should know to be better
1. Singular's Creative Benchmark Report 2026
Source: Singular (data: Q1 2025 through Q1 2026, across Social, Search and Display)
Singular published its Creative Benchmark Report for 2026, aggregated from a year of advertiser data across Social, Search and Display. The floor moved. The median advertiser now launches 13 new creatives a week, the top quartile starts at 53, and gaming's weekly counts commonly run into the hundreds.
The concentration: The top 10% of creatives capture 75% of spend in portfolios up to 100 creatives, and over 90% once portfolios pass 1,000. Bigger portfolios make it worse. In the report's words: "Running more creatives does not spread spend more evenly. It concentrates it further."
The reframe: "Creative volume on its own doesn't govern CTR, CVR, or ROI at the advertiser level." What separates similar portfolios is how fast winners get identified, and the report's repeated punchline follows: "Most advertisers could double their blended CTR by moving spend from underperforming creatives toward winners already running."
The variation math: Per NewForm's data in the report, testing one execution of a concept leaves about a 75% chance of missing the winner entirely, and two executions still miss at around 56%. It takes around 10 variations to capture 90% of winners. A winning concept only wins with a quarter to a third of its variations anyway.
The gaming edge: Gaming CVRs sit in their own category at 18.0% on Social and 17.0% on Search, which the report says makes it "the most forgiving vertical for creative testing." Speed beats polish here.
2. China's Big Bet on AI for Games
Author: Mark Ma, Naavik
Ma opens with an indie detective game on itch.io where LLM-powered suspects improvise their alibis out loud to throw you off. In China, that's no longer a novelty. AI in core gameplay has become a systemic trend there, and the production side is even further along.
The adoption depth: Tanwan Games reports over 80% AI adoption in art creation and ad targeting, and 37 Interactive produces more than 80% of its 2D art with AI while leaning on it for over 70% of ad video creation. Tencent's internal VISVISE tool speeds up character rigging by more than 8x across nearly 100 projects.
The revenue concentration: In H1 2026, AI-driven titles were 15% of China's top 100 grossing mobile games by count. They took 42.6% of that set's revenue.
The bet size: miHoYo says it will put up to ยฅ100B, roughly $14.8 billion, over three years into training systems and models. The West is polling its anxiety instead. The GDC 2026 survey found 52% of developers think generative AI hurts the industry, up from 30% a year earlier and 18% the year before that.
His warning: "Studios that spend the next few years arguing about acceptable use policies while competitors ship AI-assisted games faster and cheaper are ceding ground."
3. 172 Core Games Launched in Q1. Zero Western Winners.
Author: Matthew Emery, Gamigion (data: AppMagic)
Emery tracked all 172 core games launched in Q1 2026 and let them bake for three-plus months. Six reached a $10M+ annual run-rate, a 3.5% hit rate. Emery's summary is brutal: "All six from China, Japan or Korea. ZERO wins outside of the East."
The runaway: Last Asylum: Plague, out of China, at a $124M run-rate. That's four times the quarter's #2.
The Japan anomaly: Widen to Q4 '25 and Q1 '26 together, 170 launches across China, Japan and Korea, and Japan converted 5 of its 14 into $10M+ games. That hit rate, 35.7%, runs 3.3x Korea's and 10x China's. Nobody launches core games more efficiently right now.
The IP multiplier: On that same two-quarter set, games built on recognized IP reached the $10M+ tier at 25%, while games without it managed 4.8%. The license pays for itself.
4. The Field Guide to Stealing Ads Properly
Author: Aleksandr Dadaev, Creative Producer at Rovio (ex-Wargaming, ex-Azur Games), on Gamigion
"Competitor analysis is the bread and butter of any creative producer." Dadaev wrote down the system most teams pretend they have: finding, evaluating and remixing winning competitor ads without shipping a clone.
The search radius: Start with 5 to 10 companies in your own vertical. Then widen deliberately, into adjacent genres that share your audience's interests, non-gaming apps your players use, and a weekly 30 minutes in a general top-creatives feed to catch cross-category shifts early.
Picking the reference: Two metrics decide it, impression rate and lifespan. "If a creative has been running consistently for three months, or if it scaled aggressively within a single week, that is an undeniable signal that it's working and worth adapting."
The remix: Break the winner into its ingredients. Then substitute (turn rocks into jelly), eliminate (drop an element and see if the hook survives), or add (a reaction shot that lifts the story).
His closing shot: "We might work in a slop industry, but your analysis doesn't have to be garbage."
5. Smash Fest: 38 Games, One Mechanic From 2018
Source: Gamigion, with the clone list built alongside Gรถkhan รzmez (data: AppMagic)
The cannon-and-physics mechanic driving Smash Fest is Knock Balls, Voodoo's 2018 hypercasual hit, nearly untouched. Flow Games wrapped it in a modern puzzle economy and turned a 3-second hook into a whole game. The scoreboard: roughly 14.2M downloads, about $3.5M in IAP, and around $500K a day before it was pulled from Google Play.
Per a 2.5 Gamers analysis quoted in the article, the scaling ran about 300K downloads a day on roughly 50 creatives, with about 90% of spend on AppLovin. One excellent playable carried most of it.
The clone wave: Gamigion and Gรถkhan รzmez counted 38 Smash Fest-like games on AppMagic, together holding about 55.6M downloads and $5.1M in IAP. That set runs all the way back to Knock Balls and includes Smash Fest itself, but almost nothing underneath has any history before June 2026. The most aggressive of them, Cypher Games' Royal Smash, shipped 500 creatives in three days per the same analysis, ten times Smash Fest's volume, and sits at roughly 7.3M downloads and $620K in IAP.
The bigger idea: "Between 2017 and 2021, we didn't just make games. We built an absurdly large mechanic dataset." Hypercasual validated some of those mechanics for marketability at hundreds of millions of installs, then never built durable products on them. That graveyard is now an R&D library.
The takeaway, verbatim: "Smash Fest did NOT show us a new mechanic. It gave us a pretty damn good reason to look at old mechanics again."
6. Sensor Tower's Ad Report: The Splits Under the Headline
Author: Josh Chandley, President & COO of WildCard Games, on Deconstructor of Fun
Last issue we covered the headline of Sensor Tower's first ad monetization report, the $12 billion floor. The report carries splits that deserve their own item. They describe two industries wearing one name.
Where ads are the business: Across Sensor Tower's Jan 2025 to May 2026 window, 70.3% of India's mobile gaming revenue came from ads. Indonesia runs 55.8% and Brazil 54.7%, while Japan and South Korea sit at 10%.
The hybrid gap: Among the top hybridcasual games, the IAP-focused cohort averaged more than $83 million in total revenue over the first five months of 2026, against $21 million for the ad-focused cohort. Both models work. One buys a much bigger boat.
Where the growth went: Hypercasual Match Pair ad revenue grew $105M year over year, powered by Vita Mahjong jumping from $38M to $90M. Hypercasual Solitaire added $64M on the back of Solitaire Associations Journey. Casual Match Swap bled $155M.
The scale stat hiding in a slide: 25 billion downloads of ad-monetized games in 2025.
๐ Read on Deconstructor of Fun
7. The 95/95 Problem
Author: Aylin Yazฤฑcฤฑ, Deconstructor of Fun, profiling Emmanuel De Maistre, founder of Scenario
Buried in Yazฤฑcฤฑ's founder profile is a chart with two 95% stats on it. 95% of game developers have adopted AI in their work, per Unity's 2026 Game Development Report, and 95% of enterprise AI pilots deliver zero financial return, per an MIT study. Neither number is wrong. The gap between them is where most studios are living right now.
Where the money goes: The piece's own math: "Most studios put 80% of their AI budget into tool licenses spread across the org, 20% into some vague innovation function, and zero into rewiring how work actually gets done."
Even the AI companies: Scenario's own engineers, many of them Europeans who treat code as craft, spent months refusing to let AI coding tools into their workflow.
The widening split: In the piece's words: "Studios that say 'it doesn't work' tested one model on one use case on one day and moved on. Studios that built the AI muscle took weeks to understand the workflow, figured out which models to use for which tasks, and are now producing in days what used to take weeks or months."
The incumbents lead the adoption wave, pulling away from the challengers "because the big companies don't speak. They just do."
The headcount answer: De Maistre says none of Scenario's large studio clients cut headcount after adopting the tools. They upskilled the teams they had. Some now keep declining games alive with smaller teams and better unit economics.
๐ Read on Deconstructor of Fun
Wrap Up
Amos said something else that stayed with me longer than the operator line. From a knowledge standpoint, he figures two or three months gets a smart person everything they need to know about this job. After that, the differentiator is agency, the appetite to find the next question and act on the answer without being told.
The tools are the cheap part now. Every studio in this issue already has them. The expensive part is the person with the judgment to point them somewhere specific, and the nerve to commit budget to what comes back.
That's the bet behind Sett - the operator work runs on agents, so the hours you're paying for go into steering what to explore next and what to do with what comes back.
Go read the job description you're about to post and rethink it.
See you next time.
Yours, Jonathan "jack of all trades, master of none" 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.
