The State of Playables in UA: What Winning Teams Actually Do
Playables are eating UA strategy alive.
50x growth in 24 months. 56% of top-performing creatives now use playables, up from 30% a year ago. The top revenue-generating games are launching 64 new playables per month.
The winners moved first. Everyone else is playing catch-up.
At Sett we’re obsessed with getting closer to the truth of what winning in UA actually looks like.
This is the first of our Sett Labs series, research you can use to make better decisions about how you’re going to win in 2026 and beyond.
Less opinions. More data. Let’s dig in.
The Growth Is Undeniable
Here’s what the data shows:
October 2023: Less than 1,000 new playable creatives spotted globally. Most not even on AppLovin.
October 2025: 50,000-60,000 new playable creatives per month. Across all games. All networks.
The data is based on AppMagic and as many playables are spotted with a different in-play video, they are counted several times for the same unique playable. That being said, given we continuously count creatives that included playables in th e same methodology – we do see the trend nicely.
That’s 50x growth in two years.
When we look at the top 30 creatives by spend in October 2025, 17 of them (56%) were playables. A year ago? Only 9 (30%).
The shift is real. More UA teams are making playables core to their strategy. And they’re continuing to do it, which means performance must be driving it.
What the Top 20 Grossing Games Are Doing?
Let’s get specific. The world’s top 20 grossing games – what does their playable strategy actually look like?
The numbers:
- Median of 64 new playables launched per month (October 2025)
- Across all networks, worldwide
- These include variations, different complexity levels, new concepts
But the video-to-playable ratio is also interesting: 1 playable per 41 videos.
The leading teams push it tighter, 1 playable per 16 videos.
Why does this matter? Because playables don’t only run standalone. They run with lead-in videos. Different combinations. This gives ML-driven networks like AppLovin a ton of ammunition to systematically learn and find and optimize for pockets of high-value users.
The best teams understand this. They’re feeding the machine.
Deep Dive: What Great Actually Looks Like
Numbers are useful. But let’s drill into a single game to see how this plays out in practice.
Whiteout Survival. AppLovin. October 2025.
If you haven’t played Whiteout, you should. It’s one of the best-executed 4X games mobile has ever seen. Full-blown strategy game: base building, resource harvesting, exploration, combat. The whole deal.
Their playable strategy is even better.
Concepts vs. Variations
42 new playable ads spotted on AppLovin in October.
Here’s the breakdown:
- 15 distinct concepts – wildly different game experiences, different mechanics, different core loops
- 27 variations – not just background swaps or color changes, but smart iterations testing specific hypotheses
These aren’t the lazy variations from the old days. You know, swap a character red coat for a blue coat and call it a test.
No. These teams are methodical. Each variation asks: How do we maximize the fun of this playable concept? Or in other words, how do we create a playable that is likely to engage users deeply and for a long session?
Let’s look at examples.
The Exploit and Explore Strategy in Action
One winning concept has 11 variations. That’s 25% of all their playables, focused on exploiting a single winning idea by making the game experience more fun.
The base concept: A casualized 4X experience. Harvest resources, feed your camp, build structures, progress.
Variation 1: Adding complexity
They expanded the core loop from one resource to two. Testing whether more progression depth creates more fun – and whether players who enjoy that depth will also enjoy the full game.
Variation 2: Changing weapons
Swapped an axe for a machine gun. Testing whether faster progression (killing enemies quicker, closing the loop faster) leads to more dopamine hits and higher install intent.
Every variation has a clear hypothesis behind it.
Another Example: The Idle Tower Concept
This one’s fascinating. An idle tower building concept where you send troops – testing whether idle game lovers or “Clash’ fans will find that slice of the 4X game fun and want to try the full game.
Where did this ad run most? Cake Sort. A sorting puzzle game.
Why? Who the hell knows.
But the machine figured it out. Found a pocket of high-value players in a completely unexpected place. Harvested them. Moved on.
That’s the point.
Why This Strategy Works: The ML Connection
Here’s the thing most people miss about playables: they work because of how modern UA works.
AppLovin’s AXON. Unity’s Vector. These aren’t targeting machines where you pick an audience and serve them ads. They’re recommendation engines. Search machines. Sophisticated ML/RL systems that explore audiences, test which ones engage with different concepts, and figure out which concepts lead to high-LTV players.
You’re not controlling who sees what. The machine is.
Your job is to supply ammunition.
More diverse concepts = more shots at finding new audience pockets. Each new concept gives AXON another angle to explore. When a concept wins, you exploit it, create variations that optimize the fun and extend its performance before creative fatigue hits.
You won’t know exactly why a creative worked. The machine knows, but it can’t explain it back to you in a way that makes sense. That’s not how it works.
So you have to operate differently:
- Create diverse concepts that give the machine options
- When something wins, exploit it through smart variations
- Keep feeding it new ammunition when fatigue sets in
The top teams understand this. That’s why they’re launching 64 playables a month. That’s why they maintain ratios like 1:41.
Playables Are the New FTUE
One more insight: the best playables are pulling the FTUE higher into the funnel.
Players are starting to play your game inside the ad. The playable becomes part of their first session.
The magic happens when you layer concepts on top of each other, testing for the overlap between “players who engage with this playable” and “players who will love the full game and become high-LTV.”
Different concepts attract different types of players. When done well, each concept tests a different slice of your potential audience that might also love your full game.
That idle tower concept? It’s testing whether idle lovers or Clash fans will enjoy that aspect of the 4X experience, and be open to trying a full 4X game.
Creating concepts just for the sake of creating them misses the point. The winning teams are systematically exploring to find audiences in places you can’t expect.
The Bottom Line
The data is clear. You can’t win like the best if you don’t have a world-class playable strategy and the system to execute it.
The top revenue-generating games are launching 60+ playables per month. They’re maintaining dense video-to-playable ratios. They’re creating distinct concepts and exploiting winners through smart variations. They’re feeding ML engines with ammunition to systematically discover high-value audiences.
If you want to compete at this level, you need to create, test, and discover at the speed these teams operate.
Most UA teams can’t. The production is too slow. The hypotheses aren’t clear. The iteration loops take weeks.
That’s the gap Sett exists to close.
We’re building a data-driven playable agentic system that lets you achieve what the top 20 are doing, fast creation, smart exploration, systematic discovery of what works.
If you want in join our waitlist.
See you soon.
