Hi,

 

I like to say customer data for marketers is like chocolates for kids, they can never get enough. This has never been truer than in 2022. But usually, people think customer data should only be either historical or real-time or a combination of both. The problem with this approach is it forgets a valuable third type: predictive data.

 

You see, predictive customer data allows you to anticipate your customers’ behavior before it takes place, allowing you to act to encourage or prevent it before it happens.

 

For example, with acquisition costs exploding due to regulation changes, mobile advertising ROI is down 40% on average this year, and retaining your existing customers is more important than ever. But how to know which customers to focus on? Well, our Risk of Churn model answers exactly that question.

 

This month’s magazine will focus on an important update to our Risk of Churn predictive model, and what it means for the future of predictive analytics at Optimove. And trust me, we are rolling out new models such as predicting your top spenders and likelihood to purchase, as well as improving our existing ones.

 

Cheers,

Shai Frank, VP of Product

 

We leverage that customer data to predict a customer’s Likelihood to Convert, Future Value, Risk of Churn, and Likelihood to Reactivate, with incredible accuracy. Allowing you to predict the probability a customer will reach the next step in her lifecycle.

 

A deep dive into how we make these measurements, focused on the Future Value calculation, can be found here. But the short version is that we leverage the Customer Model that sits at the heart of Optimove and its microsegments also known as hyper granular segments,  to measure how many customers in one microsegment move to another one. The continuous analysis of these customer migrations over time allows us to predict customer behaviors. For example, the likelihood of a customer in an Active microsegment moving to a microsegment in the Churn lifecycle.

 

A year ago, we individualized customer journeys when we introduced Self-Optimizing Journeys, the ability for each unique customer to have a personalized journey that fits their historic, real-time, and predicted behavior.

 

Now, we are doing the same for our predictive analytics, starting with the Risk of Churn prediction. This means it is no longer based on microsegment migrations, but rather on the specific historical data of each customer. As a result, from now on, customers within a single microsegment can have different risks of churn. And not only that but each customer’s risk is updated daily, making sure you always have the most up-to-date prediction at your fingertips.

 

Yes, you can still see each customer’s risk of churn as a customer attribute, but we’ve also clustered customers into five groups: Critical (the top 1% of customers), High (from 90% to 99%), Medium (from 70% to 89%), Low (from 50% to 69%%), and Lowest risk (from 1% to 49%). And since these are based on percentages the customers within each cluster update daily as well.

 

While in the past you had to guesstimate which customers are 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 sending stronger promotions to them.