Provectus built a churn prediction model for Audiobooks.com from five years of historical data — listening activity, clicks, search logs, subscription renewals across 2.5 million user profiles — using ML pipelines on Amazon SageMaker and Kubeflow. At-risk customers get flagged early enough for personalised retention offers.
Benefits & outcomes
95% accuracy in churn prediction (Decision Trees, Random Forest, CatBoost).
Measurable reduction in churn and improved cost-per-circulation on audiobook titles.
Proof of concept built and deployed in five weeks.
Notes & quotes
"The churn prediction model, built by Provectus in just five weeks, allowed them to look into and analyze the specific reasons customers were about to churn."
End-to-end pipelines: data preparation, feature generation, model training and evaluation, model application.
Publisher / company: Provectus
Date of mention: 01/02/2021
Source: https://provectus.com/case-studies/user-churn-prediction-with-ai/