A Danish newspaper publisherChurn prevention in subscription products using Machine Learning

The project: Preventing churn in subscription products using machine learning

A Danish newspaper publisher wanted to reduce the subscription churn and establish an understanding of which factors and circumstances that lead to cancellations of subscriptions.

Implement helped develop a churn prediction model based on historical subscription data to predict future subscription churns.

The model was able to be integrated into the client’s IT system landscape while allowing for automatic retraining of the prediction model. When implemented, the model enabled churn predictions to become usable in both manual and automated workflows to proactively counteract subscription churns.

The impact: A machine learning churn prediction model

The machine learning model impacted the customer churn rate in their subscription products in the following ways:

• With the ability to predict 7/10 churns correctly with few false positives, the client could use the results to establish processes to counteract churns
• The machine learning models also gave insights into why customers usually churned on their subscriptions, enabling our client to increase focus in specific areas to deliver on key needs
• Results from the machine learning model was used to start strategic initiatives based on explainability of why customer churn was happening