Reduce churn and increase customer satisfaction ​through predicting super detractors​


The client was struggling to reduce customer churn among non-respondents to customer satisfaction surveys. Identifying super detractors before they churn was critical for the business.


We built a machine learning model, based on 1200+ variables analyzed for each customer, detecting potential super detractors based on their behaviors and producing key talking points for the rescue call.​

7 times more likely to identify a Super Detractor, 2 points increase in NPS given by customers


The model helped our client increase the likelihood of identifying a Super Detractor by 7 times and average NPS given by customers by 2 points.​


  1. Analysis of the in-depth behavior and attitude patterns of identified super detractors. ​
  2. Machine learning model identifying potential Super Detractors based on their behaviours, products, care interactions and usage.​ ​
  3. Daily list of top 100 customers at risk.​