marquee background
Churn forecasting
Monitor Churn Model Performance

Ensure model performance and improve customer retention

Predicting customer churn is challenging, but minimizing churn doesn’t stop once your model is in production. Managing churn model performance with ML observability helps increase overall customer satisfaction, decrease the number of customers at risk for churn, and increase customer retention.

Get Started →

Monitor churn model performance

See how Arize can help you identify and root cause model performance problems related to common model issues such as poor quality historical training data, feature/model drift, model degradation, and more.

Download Notebook →

Improve customer service and increase satisfaction through higher performing models
Monitor for market shifts to ensure your model has not grown stale
Easily track important model metrics ensure all aspects of performance don't fall through the cracks
Get ML observability in minutes.

What is Churn?

Customer churn is the percentage of customers that stop using your company’s product or service during a certain time frame.

Calculating Churn

One method to calculate a churn rate is to divide the number of customers lost during a given time interval by the number of active customers at the beginning of the period. For example, if you got 1000 customers and lost 50 last month, then your monthly churn rate is 5 percent.

Another method that organizations can define churn is through a downgrade in services so while the customer does not leave entirely, there is a reduction in revenue. To calculate churn when taking into account downgrade of services, an organization can divide the loss of monthly recurring revenue (MRR) at the beginning of the month by the total MRR.

Why Do Customers Churn?

There are many reasons why a customer may churn:

  • Poor customer service or experience. Example: Customer had an issue and their issue wasn’t followed up in a timely manner
  • Change in service options. Example: A cable provider stops including HBO in the current customer’s package
  • Dissatisfaction with service. Example: Poor reception with a particular mobile carrier
  • Market shift. Example: Many people canceled their gym memberships during a global pandemic
  • Competitors. Example: A competitor provides a discount or a superior service
  • Outgrown the need/service. Example: University student housing after graduation, most graduates leave the area upon graduating

Importance of Churn

Predicting customer churn is a very challenging task but it’s critical because the customer acquisition cost (CAC) is much higher than retaining a customer; this is very true in specific industries like tech, telecom and finance. Within American CSPs (telco), it can cost up to $315 to acquire a new customer.

Overall, it estimated that US businesses lose about $136 billion a year due to customer attrition. Churn rate is also utilized as an input to calculating a customer’s lifetime value modeling that guides the estimation of net profit contributed to the whole future relationship with a customer. Independently, it calculates the percentage of discontinuity in subscriptions by customers of a service or product within a given time frame. By utilizing machine learning to forecast churn, it allows an organization to become proactive with preventative measures instead of reacting to a cancellation notice.

ScienceSoft’s Alex Bekker also stresses the importance of machine learning for proactive churn management: “As to identifying potential churners, machine learning algorithms can do a great job here. They reveal some shared behavior patterns of those customers who have already left the company. Then, ML algorithms check the behavior of current customers against such patterns and signal if they discover potential churners.”

Recommended Resources