FinTRec: Transformer Based Unified Contextual Ads Targeting and Personalization for Financial Applications
Dwipam Katariya, Snehita Varma, Akshat Shreemali, Benjamin Wu, Kalanand Mishra, Pranab Mohanty
2025-11-21
Summary
This paper introduces FinTRec, a new system for recommending financial products to customers. It focuses on using a powerful type of artificial intelligence called transformers, which are usually used for things like language processing, but adapting them for the specific challenges of financial recommendations.
What's the problem?
Recommending financial products is tricky because people interact with companies in many different ways – online, in person, over time. This creates a lot of complex data. Also, financial companies offer many related products, and they need a system that can recommend them all effectively while also considering different business goals. Traditionally, financial companies have used simpler models because they're easier to understand and meet regulatory requirements, but these models aren't as accurate.
What's the solution?
The researchers developed FinTRec, a transformer-based system designed to handle the complexities of financial data. It's a single system that can learn from all types of customer interactions and recommend multiple products at once. They then 'fine-tuned' this system for each specific product to improve its performance. They compared FinTRec to the existing, simpler models used by the company.
Why it matters?
This work is important because it shows that more advanced AI models like transformers *can* be used effectively in the financial industry. FinTRec outperformed the older models in both simulations and real-world tests, meaning it can lead to better recommendations and potentially increased revenue. It also offers a way to streamline the recommendation process by using one system for all products, reducing costs and making things easier to manage.
Abstract
Transformer-based architectures are widely adopted in sequential recommendation systems, yet their application in Financial Services (FS) presents distinct practical and modeling challenges for real-time recommendation. These include:a) long-range user interactions (implicit and explicit) spanning both digital and physical channels generating temporally heterogeneous context, b) the presence of multiple interrelated products require coordinated models to support varied ad placements and personalized feeds, while balancing competing business goals. We propose FinTRec, a transformer-based framework that addresses these challenges and its operational objectives in FS. While tree-based models have traditionally been preferred in FS due to their explainability and alignment with regulatory requirements, our study demonstrate that FinTRec offers a viable and effective shift toward transformer-based architectures. Through historic simulation and live A/B test correlations, we show FinTRec consistently outperforms the production-grade tree-based baseline. The unified architecture, when fine-tuned for product adaptation, enables cross-product signal sharing, reduces training cost and technical debt, while improving offline performance across all products. To our knowledge, this is the first comprehensive study of unified sequential recommendation modeling in FS that addresses both technical and business considerations.