Help CenterPreempting churn involves proactive detection—identifying early warning signs before a customer shows explicit disengagement. Custom predictive models are setup and trained with the brand's specific data, and in turn analyses behavioral trends, usage patterns, and engagement signals to determine which customers may become at risk. The focus is on early intervention, that enable proactive personalised offers, retention incentives, or engagement campaigns, to steer customers away from potential churn before it fully manifests.
ℹ️ Predictive models are add-ons and require a customised setup. Contact the product team for further information.
Accessing Churned Customer Predictions
At risk customer alerts, generated by the predictive model, are available within the Inactive lifecycle stage. These alerts identify players at risk of becoming dormant and eventually churning, enabling targeted engagement strategies.
- Navigate to Inactive in the lifecycle stage section.
- Click on the menu icon in the right slider.
- Select View Predictions to access the predictive insights.
Understanding the Predictive Model Statistics
- The displayed statistics provide aggregated Key Performance Indicators (KPIs) of customers included in the predictive model.
- To the right of the statistics box, next to the people icon, the displayed number represents players who may be at risk of churning.
- Example: If the number shown is 6,136, the model predicts that 6,136 customers are likely to churn.
Engagement Options for At Risk Customers
Users can engage with predicted churned customers using two approaches:
1. One-Off Engagement
- A single campaign is launched, targeting the specific list of at risk players provided by the model.
- This is suitable for one-time incentives such as personalised offers or exclusive promotions.
2. Dynamic Campaigns
- A continuous campaign is triggered automatically whenever a new customer is flagged as at risk customers by the model.
- This ensures ongoing engagement with high-value players as they are identified.
Step-by-Step Guide to Creating a Campaign
Select Campaign Type
- Click on Target and choose One-Off Campaign or Dynamic Campaign.
Configure Campaign Details
- Assign a name to the campaign.
- Add tags for easy searchability (tagged objects can be found using the search bar at the top).
⚠️ Dynamic campaigns require both a start date and an end date.
Define Target Audience
Option 1: Target All Predicted Customers
- By default, the campaign will include all players identified in the prediction model.
Option 2: Target Customers with Consecutive Predictions
- Toggle the Consecutive Days switch to narrow the audience to customers who have been flagged as high-value for multiple consecutive days.
- Since the predictive model updates daily, this option allows users to focus on customers with consistent at risk potential.
- Pros: Increases targeting accuracy.
- Cons: If too much time passes, preemptive engagement becomes less effective.
- The dark box on the right dynamically updates to display the number of players currently being targeted.
✅ Recommendation: while the system allows for a higher number, it is not recommended to use a value higher than 3 days.
Option 3: Apply Additional Filters
- Toggle Activate Filters to refine the segment further.
- Apply additional attributes to create a more granular and customised campaign.
Choose Campaign Execution Type
- Proceed with a standard campaign setup, or
- Set up an AB/n test to test multiple variations before finalising the campaign.
By following these steps, users can efficiently create highly targeted one-off campaigns, ensuring strategic engagement with high-value customers while maintaining flexibility in audience selection.
✅ Suggested reads: