Business Situation
- An insurance company sought to optimize its customer acquisition strategy and increase wallet share from existing customers.
- The company faced challenges in effectively targeting potential customers and identifying early adopters for its new product offerings.
SGA Approach
Technology
- Leveraged AWS SageMaker for model deployment and batch inferencing.
- Implemented XGBoost classification ML model with Bayesian optimization.
AI
- Created 500+ predictive features using advanced feature engineering techniques.
- Applied mathematical transformations to improve feature information quality after utilizing SHAP and LIME frameworks for explainable ML.
Data
- Integrated and processed diverse data sources: first-party, third-party, and survey data.
- Handled missing values, outliers, and imputations to ensure data quality.
- Implemented rules-based classifications for loyal accounts and spend categories.
Key Takeaways
- Developing a sophisticated ML-based targeting model helped predict purchase propensity across various insurance offerings.
- Implementing explainable AI (XAI) frameworks, like SHAP and LIME, provided higher transparency in decision-making.
- 30% increase in wallet share from low spenders led to $25 million in additional revenue.
- Creating a comprehensive data integration strategy enabled the combination of first-party, third-party, and survey data.
- Identifying potential early adopters of product offerings increased the conversion rate by 9%.
