Granular segments increase conversion by 4x-9x for bank

One of Asia’s largest retail banks wanted to build AI-powered decisioning capabilities to  execute omnichannel personalized experiences for its 45 million customers. Their legacy, on-premises data was located in various databases and data silos across the organization, and so the challenge was twofold: 

• First, in order to action a number of use cases at scale, the bank wanted to build a centralized data universe with next-gen on-cloud infrastructure.  
• Once that’s implemented, the bank required analytics technology that could be operated on more than 10 marketing channels, while discovering and managing thousands of customer segments every day. 

For the first use case, the marketing team wanted to combat churn by identifying high-risk customers. The churn rate for bank deposits was 15-20% p.a. for existing customers. 

SparkBeyond led the migration of > 40 TB of complex and disparate data onto the cloud, while managing the monthly ingestion of 40GB of data, including 1.5 billion rows of data to score per month. 

The data fields included call center data, customer applications, transactional data, fraud data, cross-sell attempts, and others. Once integrated with SparkBeyond, the platform’s data-type detection capability used geospatial data, customer complaints, bureau data, loan rejections, and relationship manager performance data to create highly granular segments for operationalization.

The platform’s automated data ingestion reduced time spent on data wrangling and preparation by 35%, while its intuitive model building led to a 30% increase in quality time spent on applying domain expertise, interpreting features, discussing marketing interventions.

SparkBeyond’s powerful, bias-free, and rapid hypothesis generation (millions discovered in a few hours) led to significant improvements in the quality of AA insights. Leveraging SparkBeyond enabled the bank to action a number of use cases at scale, including scoring approximately 1 billion records per month.

For example, the customer retention use case surfaced >5k intelligent insights and ~300 niche customer segments through rapid data engineering (>20+TB data auto-fed into models). This helped the bank service staff to have meaningful conversations with end-customers as they have more insight into what the customer might need. By ensuring the bank’s relevance to existing and potential customers, reducing spam, and increasing the satisfaction level of customers when banking online or when calling support – the bank saw a 4x-9x sales increase across a range of its products.

What’s more, by leveraging SparkBeyond on the cloud, the bank can now  generate granular insights without exposing personal data – a crucial capability that accentuates the relationship of trust between a bank and its customers.

‍