The company's CRO and his team was looking to employ a more sophisticated data driven approach towards understanding and predicting the urgent customer churn challenge. However, the in-house ML resource was pre-occupied on other projects and it would take months before the limited ML resources in the company could undertake the urgent churn prediction problem.

One key data set for churn prediction is product usage data and the company’s own app’s wasn’t instrumented. This was a road block to build a useful churn prediction model. That’s when they turned to GrowthSimple when they heard that GrowthSimple has an innovative approach to collecting product usage data from back end api logs.

The Solution

The product analytics tools were good for understanding product related metrics but aren’t setup for ingesting other data sets such as payment transaction, user demographic etc. to build an actionable churn model. It took GrowthSimple, team less than a week and one call with one backend engineering resource to integrate product usage, user demographic, transaction and label data and answer the following questions:

What is the most important churn challenge?

7 day

30 day

90 day

Data exploration uncovered that the 30 day was where the most significant churn was occuring. For the 30 day window, we provided users grouped and ranked into three horizon windows - 1 week, 2 weeks and 3 weeks.

The Result

GrowthSimple platform came up with a 106% churn prediction boost, which was validated against actual churned users by the customers' BI team data analyst. The GrowthSimple platform provides a list of churned users ranked by churn propensity, weekly and pushes them directly into Salesforce Pardot Object used by the Customer Marketing team.

GrowthSimple’s platform allowed the BI team to not worry about maintaining or tune the model as more data becomes available and at a significantly lower cost. ‍Most notably, GrowthSimple empowered its existing BI team with the capability of building state of the art AI predictive analytics models, without the need for an in-house ML Engineering team and they are now thinking of optimizing the predictive model by utilizing UTM data, customer intent data (through exit surveys) and third party data sets.

Get Started Now

Learn how SaaS and Subscription e-commerce companies significantly scale revenue.