Cutting customer churn by 10% in insurance with ML 

Kubrick helped a leading insurance firm deliver prevention marketing strategies that mitigated customer churn and unlocked revenue. 

1. 01

   The challenge

   In a hyper-competitive market, with customers easily switching to take advantage of new offers, churn was a challenge to our client’s sustainable revenue growth.​ 

   Marketing lacked the analytics capabilities to accurately predict at-risk customers and avoid wasted campaign efforts or ineffective messaging strategies. ​ 

   To improve their marketing strategy with more accurate targeting and messaging to drive valuable retention, the client looked to Kubrick to use machine learning to identify the customers most at risk of churn. 

2. 02

   The approach

   Kubrick developed a churn prediction model, using advanced machine learning techniques, optimized with time-series validating and fine-tuning. ​ 

   Trained using an extensive set of historic customer data (85,000 churn events and 345,000 non-churning observations), the model determines risk indicators, behaviors, and demographics to categorize customers into levels of churn risk. ​ 

3. 03

   The impact

   • Enhanced churn prediction: By categorizing customers within specific timeframes, the tool indicated the customers who were 50-75% most likely to churn within 6 months for effective intervention. With a model accuracy of 70-80% in initial trials, the tool delivered a 10% reduction in customer churn at the lowest bound of possibility. ​ 
   • Optimized marketing cost and attributable ROI: Improved targeting and messaging on at-risk customers reduced wasted costs and provided ROI on mitigated churn, exceeding $220,000 for a 10% churn reduction.​ 
   • End user adoption: The tool had a 100% adoption rate in the marketing team, driving YoY ROI and expanding the use case into other areas of Customer Lifetime Value attribution to continue driving optimized marketing that delivered revenue growth.​