Mosaic Data Science developed a predictive model for a major enterprise software provider experiencing declining maintenance and service renewals in North America. To combat this, the team designed a proof-of-concept model using decision trees, trained on data including service calls and software download records. This model aimed to predict customer churn up to 18 months before contract renewal. Decision trees were chosen for their interpretability, allowing business stakeholders to understand the results. The models achieved over 70% accuracy, effectively highlighting at-risk customers. Key churn predictors included service call volume and history of prior cancellations, enabling proactive customer engagement and improved retention strategies.

Join for free to read