Introduction to Predictive Modelling

Predictive modelling leverages historical customer behavior and a blend of statistical models to forecast future player actions. This approach enables the CRM to anticipate player traits and movements before they actually occur, allowing for proactive engagement and tailored marketing strategies.

Predictive modeling alerts and campaigns can be accessed directly by clicking on the lifecycle stages within the CRM dashboard. When any alerts are available, the respective lifecycle stage is highlighted in red upon accessing the dashboard.

ℹ️ Predictive models are add-ons and require a customised setup. Contact the product team for further information.

Predictive Modelling Alerts Overview

Predictive modelling alerts deliver timely insights into customer behavior and are seamlessly integrated within the lifecycle stages on the dashboard. These alerts enable organisations to identify and engage specifically with the customers flagged by the predictive model, facilitating highly targeted campaigns.

Accessing Predictive Alerts

• VIP Predictions:
  Accessible from the New Depositors lifecycle stage on the dashboard, the VIP model highlights players with the potential to become high-value customers.

• Inactive Predictions:
  Available in the Active lifecycle stage on the dashboard, these predictions identify players who may be at risk of becoming inactive.

• Churned Predictions:
  Located within the Inactive lifecycle stage, these alerts focus on customers who are likely to churn.

Viewing and Interpreting Predictions

Within the lifecycle stage slider, clicking on View Predictions reveals a comprehensive display of key performance indicators (KPIs) and statistics, including:

• KPI Metrics:
  Various performance metrics related to the predicted customer behavior.

• Total Customer Count:
  The number of customers included in the prediction is clearly indicated, enabling quick assessment of the scope.

Initiating Campaigns from Predictions

Based on the predictive insights, organisations can act immediately by targeting the identified customer segments through tailored campaigns:

• One-Off Campaigns:
  Initiate an immediate, single-use campaign to engage the customers flagged by the model.

• Dynamic Campaigns:
  Set up a dynamic campaign that automatically launches whenever a customer meets the prediction criteria, ensuring continuous engagement.

VIP Predictions

Objective:
The CRM employs predictive modelling to identify players with the potential to become VIPs—typically defined as the top 5% high-rollers in the database.

Methodology:

• Data Analysis: The model examines player activity, wagering patterns, and deposit behaviors.
• Lifecycle Stage: VIP predictions are primarily generated during the New Depositors lifecycle stage.

Benefits:
Early identification of potential VIPs allows organisations to monitor these players closely and implement strategies to nurture their loyalty and engagement.

Churn Detection

Objective:
Churn detection aims to identify players at risk of becoming inactive—and eventually churning—up to 14 days before inactivity occurs.

Two-Level Prediction Model:

1. Active Player Analysis:

   • Approach: The model analyses the behaviors of active players to detect early signs of disengagement.
   • Indicators: Specific traits or changes in activity patterns that may signal an impending shift towards inactivity.

2. Inactive Player Analysis:

   • Approach: For players already inactive, the model evaluates their responses to campaigns, engagement levels, and overall activity.
   • Outcome: The analysis predicts whether these players are likely to return to an active status or progress toward dormancy and eventual churn.

Operational Efficiency:

• Daily Reviews: Predictions are automatically reviewed and updated on a daily basis, ensuring ongoing accuracy.
• High Accuracy: With an accuracy rate of up to 79%, the model is designed to favour false positives—flagging potential at-risk players even when the probability is around 50/50—to ensure that no potential churn candidate is overlooked.

This refined predictive modelling approach helps organisations optimise their player engagement strategies by preemptively addressing potential issues and tailoring interventions to both high-value and at-risk players.

✅ Suggested read: Data Analysis on Model Outcomes