Skip to main content

Understanding Risk of Churn

Sophie Grobman
  • Updated

A company’s ability to prevent customers from churning is directly linked to its profitability. Identifying which customers are at risk of churn tells us when we should send win-back offers and encourage a return to activity. 

With the Risk of Churn attribute, you will be able to: 
  • Easily target customers based on their risk level to churn 
  • Allocate promotional budgets in the most optimal way 
  • Take proactive measures to combat churn in real-time 

Understanding Risk of Churn

Risk of Churn is a predictive metric that measures the likelihood of churn for customers in live lifecycle stages (e.g., New, Active, or Reactivated).

Our machine learning models analyze historical data to identify common behaviors, attributes, or signals of churned customers. These can include visit or purchase frequency, recency, average transaction amounts, and demographic factors.

The score is updated daily and reflects the likelihood of a customer churning within the next two periods. In other words, it’s a dynamic, real-time prediction that helps you proactively manage churn risk across your live customer base.

 

How to Use

The risk of Churn is reported in the Customer Model and can be used to define target groups in the Customer Explorer and Manage Target Groups pages. 

Optimove will proactively segment clients based on their likelihood of churning into five risk levels: Critical, High, Medium, Low, and Lowest. These groups are measured as a percentile: meaning each group is compared to the other groups from the same set of customers.  
 
The example below shows 95 customers at Critical Risk, with the highest probability to churn, ranging from 99%-100%. Whereas, at the Lowest Risk level, we see 4,771 customers, with the lowest probability to churn ranging from 0-52%. This tells us, that the customer base of this company is healthy, as the majority of the customer base is at a low likelihood to churn.
 
 
 

Can I customize my Risk of Churn model?

Each business has its own trends, seasonality, and characteristics which may affect churn percentages and retention rates. As a result, each marketer classifies which customers are at risk of churn and how they should target these customers differently.
 
Some may see ‘at risk’ customers as those who have a probability to churn greater than 70%, whereas others define those ‘at risk’ as the top 15% of their customers in terms of their probability to churn.
 
To better adapt your marketing strategy to improve retention, you might want to harness the model’s power and use it differently than our default recipe. 
 
We, therefore, enable you to set your own threshold on how customers are classified in Risk of Churn using customer attributes. This allows you to then identify customers that are at risk, define a strategy of churn prevention, and target users based on the values in this attribute.
 
To do this, you can create a new Target Group using one of these three customer attributes:
 

a) Churn Probability Score

This attribute allows you to segment live lifecycle stage customers based on their predicted risk of churning. The score reflects the probability that a customer will become inactive within the model’s time frame.

You can adjust the score range to build targeted segments for proactive retention.

Examples:

  • Churn Probability Score ≥ 0.7 will target live customers with a 70% or higher chance of churning soon.

  • Churn Probability Score < 0.3 can identify low-risk customers, ideal for loyalty or upsell campaigns.

b) Rank in Churn Probability

This attribute ranks live lifecycle stage customers into percentiles based on their Churn Probability score. It helps you prioritize retention efforts relative to the rest of your customer base within these stages.

 
For example, if you consider High-risk customers in the Risk of Churn model to be customers in the top 5%, you can create a target group of customers whose Percentile IN probability TO migrate TO churn is larger than or equal to 95.
 
This will then create a target group of customers that have a higher probability to churn than the other 95% of the live customers.  
 
 

Use this percentile view to apply tiered strategies such as tailored offers based on risk level.

 

c) Rank in LCS of Churn Probability

This attribute ranks customers by Churn Risk within their Lifecycle Stage (live stages only). It is especially useful when churn-prone customers are concentrated in one of these stages and you want to ensure balanced coverage across all relevant stages.

Examples:

  • Rank in LCS of Churn Risk = 100 will segment the top 1% of at-risk customers within each lifecycle stage:

    • Top 1% of churn risk among Active customers

    • Top 1% among New customers

    • Top 1% among Reactivated customers

Using this attribute ensures retention strategies are aligned across the live lifecycle stages supported by this model.

 
 

 

Be There Before They Churn

While in the past, you had to guesstimate which customers were at critical risk, today, you can just select the “Critical” option under the Risk of Churn selection criteria. Making it easier than ever to, for example, identify customers that need a more aggressive retention strategy and send stronger promotions to them. Specifically, we recommend these two approaches when using the Risk of Churn attribute:  

  1. Proactively prevent churn for the segments that need it. For example, if you are launching a promotion campaign, instead of sending the same promotion (10% off) to all clients, you can offer higher or lower discounts, based on the client’s likelihood to churn. This way, you allocate your promotional budget in the most optimal way. 
  2. Measure the risk of churn per acquisition channel to optimize your acquisition campaigns. For example, you might be optimizing acquisition channels with lower CAC, but at the end of the day, these “cheap” acquisition channels will yield high-risk customers. The Risk of Churn metric will allow you to assess the real and long-term value of your acquisition strategy.  
 

Related articles