Business Impacts
Increased cost savings due to marketing optimization
Improved efficiency of the model by 25%
Customer Key Facts
- Country : United States
- Size : 17,000+ Employees
- Industry : Banking
- About : The client provides banking, investment, insurance
and mortgage financial services to over 3.6 million
consumers, business and government agencies.
Leveraging Data from Various Sources to Build Machine Learning Models
The client is a multi-state community-focused bank serving in New York, Maryland, and other states in the US and provides banking, investment, insurance, and mortgage services to its clientele. The client wanted to leverage data from multiple sources such as clickstream, account, and transaction history to build machine learning models. It also wanted to use Google Ads Customer Match API to reach out to users with high conversion propensity scores.
Challenges
- The client provides banking, investment, insurance, and mortgage financial services to over 3.6 million consumers, businesses, and government agencies.
- ~ 0.02% conversion rate for mortgage purchase, creating high data imbalances
- Including limited features in the model for fair lending
Technologies Used
Google BigQuery
Google AI Platform
Google Compute Engine
Google Cloud Platform
Google Ads
Solution
Quantiphi helped in mapping the probability scores against different customer identifiers such as user login ID and WebID so that model outputs can be used to reach out to customers on different channels. Quantiphi created profiles and segments of people seeking mortgage services from banks to allow enhanced retargeting and customer experience.
Result
- Increased cost savings due to optimized marketing
- Up to 25% improvement in efficiency of the model
