Churn prediction for Audiobooks.com

Amazon, Provectus, RBmediaBuilt by: vendor working with a publisherStatus: LiveYear: reported in 2026

Overview

Provectus built a churn prediction model for Audiobooks.com, the digital audiobook platform of RBmedia, which the case study calls the world’s largest audiobook publisher. Customers can switch platforms easily, churn affects Audiobooks.com’s main metric of cost per circulation, and its earlier churn work had been exploratory.

Provectus first set up an evaluation framework and then trained models on 2.5 million user profiles with five years of listening activity, search and recommendation logs, subscription renewals and churn events, using Amazon SageMaker for training, evaluation and scoring. The final model uses 100 features covering subscription behaviour, recency of activity, search frequency and credit usage, shows the factors behind each at-risk user’s prediction, and is owned and run by Audiobooks.com.

Benefits and outcomes

  • Provectus reports churn prediction accuracy of 95% or more across test scenarios, with several model variants reaching 98% to 99%.
  • Provectus reports that the model went from prototype to production in five weeks.
  • Provectus reports that marketing can target at-risk users and avoid contacting users who were never at risk.

Sources

This entry was written with the help of AI, from the sources listed. Please check the sources before relying on it, and let us know if anything needs correcting.