50% Higher Campaign Success with AI-Driven Customer Behavior Modeling

As part of its digital strategy, Raiffeisenbank Czechia set out to leverage customer and transactional data to deliver more personalized offers, improve campaign targeting, and increase the conversion rate of its large-scale consumer loan marketing efforts.

Achieving this required a sophisticated solution capable of performing highly complex calculations on hundreds of millions of transactions and identifying relationships within the customer network—a challenge beyond the capabilities of traditional relational databases and standard business intelligence tools.

Challenges

The solution had to be deployed within the bank’s existing big data platform based on Hadoop technology. Additionally, the campaign outputs needed to be technically integrated with the bank’s campaign management tools and aligned with the client advocacy and contact policy processes.

Solution & results

Profinit’s AcceptAI, now part of the Amdocs portfolio, was chosen as the core solution. AcceptAI models customer behavior using transactional data and applies machine learning to calculate propensity scores for individual customers across different financial products.

To handle the computational complexity, the system employs parallel processing via Apache Spark on the Hadoop platform, enabling the processing of several years of transaction data in a matter of minutes.

The resulting propensity scores are made available in the bank’s Data Warehouse (DWH) for use by campaign management tools. This allows the bank to rank and segment customers effectively, select the optimal communication channels, and ensure compliance with outreach policy rules.