Case Studies

Mitigating Churn for an Automotive Manufacturer’s Subscription Service

Posted by | Fuld & Company

An automotive manufacturer’s aftermarket subscription service was experiencing high churn, and management wanted to review the situation and take corrective measures.


To identify churn drivers and create a predictive model to flag customers at risk.


  • Understood the client’s business model and analyzed historical data, including service delivery, feedback, and complaint data.
  • Created train and test data using currently active and churned customers.
  • Leveraged Random Forest and Extra Trees models to identify the drivers of customer churn.
  • Created a neural network-based classification model to flag customers at risk; created sales plays that recommended products and discounts that sales agents could propose to customers; redesigned agent incentives to leverage the model’s recommendations and provide feedback to improve the sales plays.


  • Churn rate reduced by 13.6%.
  • The company can identify customers at risk and take timely actions to mitigate churn risk.
  • Created roadmap to enable sales plays experiments among customers via sales call centers

Algorithms Used: Random Forest, Extra Tress, ANN


Tags: , , ,

Related Resources

Read More

Global medical equipment company discovers a new competitor…and a new market

A leading medical equipment manufacturer was concerned about a new entrant that had moved into one of its high-value, diagnostic […]

Read More

Identifying Drug Repurposing Collaboration Partners

A top pharma company actively working in the drug repurposing sector wanted to expand its portfolio by adding new indications […]

Read More

Identifying the most effective prophylactic/therapeutic oral anti-inflammatory agents

A global FMCG leader sought to expand their current oral care product range with a line of products that had […]

Subscribe to our mailing list for our latest updates:

Wednesday, February 15

10:30 ET | 15:30 GMT

Join us to learn about:

  • Cultural nuance issues
  • Methods to mitigate bias
  • Optimizing survey design
  • Interpreting global results