Churn, as scary as it is real, means the end of a relationship. It affects all businesses at some point and no one, not even the most successful retailers in the world, is excluded.

Churn is a fact and you have to work very hard within every corner of your business to prevent and reduce customer churn. One very effective way of managing this is to measure and act on predictive churn scores.

Voyado keeps an eye on each customer’s overall engagement to make an assessment whether this customer shows signs of declining interest in your business. Looking at things such as opening rates, click rates, purchase frequency, etc.,. Churn score is the probability (0-1) that a customer will fall below a certain limit value in their purchase frequency within a certain time period.

The time period and limit value are unique to each Voyado customer, based on how the normal purchase pattern looks like according to the model.

The churn scoring predicts behavior with a certain margin of error. This means that the events that the scoring are forecasting may not yet have happened. Using this type of values means that you can keep an eye on those customers which are predicted to have a high risk for churn.

Score Description
0.0 - 0.50 Active (Low churn score)
0.51 - 0.75 Declining (Medium churn score)
0.76 - 1.00 Churning (High churn score)

How to take action on predictive values

If a customer is predicted to have a high churn, it is very likely that he/she does not open your emails often and does not have an overall interest in your brand anymore. For this reason, it is recommended that you test different channels to approach the customer to see which channel works. Also keep in mind that you will probably not rescue the majority of your churners, but pushing an occasional extra visit to your site or store will add up in the long run.

Prevent churn as early as possible

A possible use case for preventing churn is to combine the predictive churn score with recent behavior. Set up an automation that fetches possible churners and process them with offers relevant for their profile. Here is an example:


  • At the end of every month, check all customers that made a purchase during the last month.
  • Now check the churn scoring. Is this customer flagged as a potential churner? If so, this could in fact be the last purchase in quite a while.
  • How do you rank this customer? Is this a customer that usually have a large order value or receipt? Then this is a customer you might want to spend some money on to get one or several more purchases from.
  • If this is a customer that did not have a large order value in the past, you might want to try and keep them around but not spend a lot of your budget on trying to do so.
  • Create a randomized split and try different channels and measure the response.


Next step: Read about CLV - Customer Lifetime Value

Was this article helpful?
0 out of 0 found this helpful



Please sign in to leave a comment.