Time to Count: How We Build Predictive Analytics in myTracker

What the talk is about

Artem Khapkin explains how predictive analytics and LTV forecasting for games and apps are built in the myTracker mobile and web analytics platform. He covers why marketing teams need early revenue forecasts across in-app purchases, subscriptions and ad monetization, and how predictive models help evaluate user acquisition payback within the first days of a cohort's lifecycle. He walks through building custom ensemble and gradient-boosting models, handling cold start for newly launched apps via clustering of similar projects, and combining revenue data across multiple sources and ad mediation platforms. Useful for product and marketing analysts, data scientists and mobile UA managers.