Anytime a client cuts ties, you encounter the negative consequence of customer churn. While some churn is a natural component of any company, a vast churn rate can cripple any organization's growth. Quaartz can help recognize customers liable to drop so that the sales team can take timely actions to retain them.



Project Duration and Effort

Two days

Type of Prediction

Binary Classification

Customer Benefits

  1. Identification of customers likely to leave the company with a 73% accuracy.
  2. It gives the sales team the means to prioritize actions to retain customers based on their churn likelihood.
  3. Speed, once the tool is in place, Quaartz’ analysis takes a few minutes.

Problem to solve

According to PWC, 32% of all customers would quit doing business with a brand they liked after a single unsatisfactory experience. And 73% of U.S. consumers declare that customer experience is a significant factor in their purchasing decision. It is so powerful that even if they love a company or product, 59% will still walk away after several bad experiences, and 17% will walk away after just one bad experience. A Mckinsey study shows that happy customers are more likely to upgrade or add services and are less prone to canceling. Confronted with poor customer service, 20% of consumers would complain publicly via social media says NewVoiceMedia.
In a nutshell, it is of prime importance to keep customers happy and anticipate the red zone into which a customer might walk out the door and display his/her dissatisfaction publicly on social media.

We want to see whether Quaartz can help identify which customers are likely to leave soon. We use a dataset containing customer-related data, which includes the following information:

  1. Customer gender
  2. Whether the customer is a senior citizen
  3. Whether the customer has a partner
  4. Whether the customer has dependents
  5. Number of months with the company
  6. Has a phone service
  7. Has multiple lines
  8. Has an internet service provider
  9. Has online security
  10. Has online backup
  11. Has device protection
  12. Has technical support
  13. Has streaming TV
  14. Has streaming movies
  15. The contract term
  16. Has paperless billing
  17. The payment method used
  18. The amount of monthly charges


We want to estimate who is likely to leave the company. And we would like to understand the characteristics of those customers likely to leave,i.e., how to recognize them before leaving. It would give us the tools to retain them proactively.


  • Estimate the likelihood of a customer leaving.
  • Characterize the customers with the highest probability.
  • Give the means to the sales team to take corrective actions.


This issue raises the following marketing question:

Can we identify with a good reliability a customer likely to leave?


We have run Quaartz on the above-described dataset asking the tool to identify customers likely to churn. Quaartz predictive models’ results reach a 73% accuracy and an 81% AUC based on this real dataset. Avoidable consumer switching costs U.S. companies $136.8 billion per year, according to CallMiner.

Quaartz has selected the following four main criteria to make its decisions out of the twenty available in the dataset:

  • The number of months with the company, with a weight of 38% on Quaartz’ decision,
  • If the customer has internet security, with a weight of 22% on Quaartz’ decision,
  • Contract term, with a weight of 20% on Quaartz’ decision,
  • And a specific internet service provider, with a weight of 19% on Quaartz’ decision.

Consider two companies: company A and company B, with both an annual revenue of 20 million. Assume they are both growing at the same rate: 50%. Company A has a yearly churn of 5%, and company B has an annual churn of 15%. After five years, company A’s revenue is 90 million, and company B’s income is 60 million. This example shows that avoiding churn has a significant impact on revenue.
Quaartz helps customer satisfaction teams improve their performance and retain more customers.

Customer Benefits

In two days, the customer satisfaction team gained a competitive edge against their competitors through improved churn prevention:

  1. Estimate the percentage of customers likely to churn in minutes with a 73% accuracy.
  2. Identify which characteristics these customers have in common.
  3. Obtain an immediate “what-if” analysis indicating the dependency between their features and their likelihood of leaving.