The document describes a project to predict customer churn and upgrade opportunities using product usage data. A random forest model was developed to classify customers as either churned or upgraded and not churned within one year. The model achieved AUCs of 0.764 and 0.882 respectively. The company was able to gain insights into important factors like early user activity signals and increase monthly recurring revenue over 100% through targeting upgrades. Further improvements to the model were proposed, including additional product and user data.