Predict Customer Churn

The Churn Prediction application classifies whether a customer has churned or not. For customers who haven’t churned, it predicts if a customer is expected to churn (expected to reduce or stop purchasing) and the churn timeframe (when the churn is expected to happen).

This application provides:
  • Automated insight into at-risk users:
    • Can provide visibility to stakeholders on where to focus resources to limit churn to the most at-risk users
    • Allows business to segment at-risk user groups and build tailored retention efforts
  • Early warning of potential revenue impacts:
    • Predict impact on revenue of losing at-risk users within time period, allowing businesses to take preemptive actions
    • Help business quantify the hidden costs of churn further
    • Cost of acquiring new users
    • Potential negative impact of brand reputation
    • New user onboarding costs

This application uses your sales history that includes the sales order, status, customer address, currency, item, and payment method details to predict the churn. If the gap from today to the last purchase in days falls in the 90th percentile of all historical delays, the application identifies it as a churn automatically. You can manually specify the number of months that constitute a churn (months since last customer transaction date). If last transaction date is outside of this window, the application identifies it as a churn.

Ensure to activate the Sales functional area prior to activating this application.

  1. Select the toggle for Enable NetSuite Predictive Applications in Pipeline Features under the Preview Features tab on the Enable Features page. See Make Preview Features Available.

    Pipeline Features in Preview Features tab on the Enable Features page

  2. On the Generally Available Features tab, select the toggle for NetSuite Churn Prediction under Functional Areas.

    NetSuite Churn Prediction under Functional Areas in the Generally Available Features tab on the Enable Features page

  3. Navigate to Data Configuration, under Applications, click NetSuite Predictive Applications.
  4. On the NetSuite Predictive Applications page, click Create, in step 1 of the NetSuite Predictive Applications wizard, select Oracle NetSuite Predictive Analytics Applications in Offerings, select Churn Prediction in Functional Area, and click Next.

    NetSuite Predictive Applications page for Churn Prediction functional area

  5. In step 2 of the wizard, select the parameters to use for the functional area that you've selected and click Next:
    • Months of transaction data to use – The past number of months of training data required for training a prediction model. For example, if you specify 72, the application uses the past 72 months of data for predicting the churn.
    • Churn Identification Method – There are 2 methods of determining churn. One is automatic and the other is manual. In the case of manual method, if you haven’t purchased in past certain months, the application considers those number of months as churn. For the auto method, the application identifies churn if you didn’t purchase within a timeframe determined by an algorithm.
    • Expected churn rate (only for auto method) – This parameter controls the percentage of churn.
    • Absence duration (only for manual method) – Number of months without purchases.

    Parameters for the Churn Prediction functional area

  6. In step 3 of the wizard, review your selections for the functional area activation and click Next.
  7. In step 4 of the wizard, select when to initiate the functional area activation and click Finish.