Decrease customer churn rate through predicting who are the most likely churners​


The client was struggling to simultaneously identify the customers, who are the most critical churners while also sustaining the highest return on investment from marketing campaigns.​


We first applied a logit model to understand what were the most impactful drivers of churn. Additionally, the effect of the drivers was quantified. ​Next, we used extreme gradient boosted trees as the approach for predicting the churn on an individual level. ​

Reduce churn and increase customer satisfaction through predicting super detractors​


We helped our client to efficiently spend the marketing budget and target efforts only at customers that were classified as part of the top percentile churners, reducing churn by 20% among the likely churners.​


  1. A list of significant differences between the drivers of churn.​
  2. A list of suggestions for effective marketing campaign targeting. ​
  3. A churn score for each of the active customers of the client, allowing them to identify the ones that are most likely to churn in the upcoming months​.