Gaming
Last-touch attribution fails when players discover games through layered campaigns. You need ROAS distributions, not point estimates.
Mobile gaming is one of the most data-rich, fast-moving industries in the world. Standard last-touch attribution fails when players discover games through layered campaigns across social, programmatic, and rewarded video. Bayesian MMM gives you ROAS distributions with credible intervals, so you can see which channels have high return and which have high uncertainty.
We don't give you a point ROAS. We give you a distribution. You can see which channels have high expected return but also high uncertainty.
Work in This Space
Supercell
Hierarchical Bayesian MMM for Mobile UA
Hierarchical Bayesian MMM with market-level structure, ROAS and budget optimization for one of the world’s largest mobile gaming studios.
Appodeal
Mobile Ad Attribution MMM
Custom Bayesian MMM for mobile ad mediation with adstock, saturation, and hierarchical structure. ROAS estimation with credible intervals per channel.
Trusted By
- Supercell
- Appodeal
- Keywords Studios
- Game Data Pros
What We Solve
Media Mix Modeling for User Acquisition
Hierarchical Bayesian MMM across acquisition channels with adstock, saturation, and time-varying intercept. Posterior ROAS with credible intervals and budget optimization.
Player Lifetime Value
Probabilistic LTV modeling via Pareto/NBD and BG/NBD variants. Full LTV distributions for marketing managers, PE analysts, and CFOs.
A/B Testing at Scale
Hundreds of simultaneous experiments for matchmaking, pricing, UX, and level design. Bayesian frameworks handle multiple comparisons without alpha inflation.
Synthetic Player Research
LLMs as synthetic respondents for evaluating ad creative. Statistical validation comparing LLM vs. human score distributions for faster, cheaper creative testing.
Why Bayesian
Why Bayesian for Gaming?
- Uncertainty-native ROAS — Full posterior distributions over channel returns, not point estimates. See which channels have high return and high uncertainty before committing budget.
- Principled A/B testing at scale — Multiple simultaneous comparisons without inflating false positive rates. Early stopping with Bayesian decision rules saves time and money.
- Domain expertise built in — Our partner Luca Fiaschi was Chief Data & AI Officer at MistPlay, bringing direct domain expertise in UA efficiency and player engagement modeling.
“We want everything that Christian showed. The Bayesian approach to MMM is exactly what our UA team needs.”