Metica's best product yet, SmartFloors: ad monetisation that runs itself
Shipped in
2025 – Ongoing
My role
Founding product designer
Team
Metica's whole team
Every previous version of Metica asked the customer to configure something. SmartFloors asked them to configure nothing. It optimises bid floors on top of AppLovin MAX, modelling the real value of each player and adapting the auction in real time. The customer integrates once, then leaves it to run. The contextual bandits that started life inside the offers platform finally had a single, high-value problem to point at. This is the product Metica is built on today.
The impact
81+ games integrated, expanding monthly
~11% revenue-weighted ARPDAU lift across the portfolio
Integration reduced from weeks to hours
The most black-boxed product we ever built
Customers had been clear across every prior product: integration was painful, and they didn't want to manage a hundred parameters. They wanted something that worked and a result they could look at. SmartFloors took that literally. Almost everything the customer used to touch moved under the surface.
The smallest effort for integration so far
We built a dedicated SDK team, now five people, focused on one thing: making integration take hours, not weeks. Once it's in, Metica takes over, combining the bandit-driven optimisation with domain expertise and data held across the customer base. From the studio's side, there is nothing left to configure.
Prototype of the integration testing flow for a game studio developer. Check it out here
Plug it in and leave it to run
The product is a combination of machine learning and accumulated domain knowledge, working continuously beneath the surface. The customer plugs it in, then steps away. What's left on their side is a reporting page, and a support team that answers questions and provides deeper analysis when they want it.
When the design problem becomes the result, not the controls
With configuration gone, the design challenge moved. The work now is helping customers read and trust outcomes they didn't tune themselves. Across 81+ integrated games and growing every month, the question stopped being "how do I set this up" and became "what is this telling me, and can I believe it." That problem is where the product is evolving now.
Read more about it in this next article about building a Slack AI teammate.
