Personalized Banking Products Recommendation

Recommending the best product to banking customers based on transaction data

What to offer next and why?

Our partner, one of the largest Canadian banks, would like to proactively offer its customers products and services to improve client satisfaction and increase retention. The bank has rich data on client profile, credit card and debit card transactions, digital interactions, and call center interactions.

One of the key challenges the bank faced in the past with best next-offer solutions was that the recommendations were typically the output of a blackbox model that could seem out of place and couldn't be explained. As a result, a banker on the phone or sitting next to the customer would have no context and would be unable to engage with the customer on how the offer is a good fit. Our goal was to provide a meaningful narrative in addition to the predictions.

Client-journey-driven segmentation

We used Optimal Classification Trees to predict which products the customer is likely to engage with in the next 3 months.
  • Rich history from transaction data

    Based on detailed transaction data from various sources, the algorithm automatically combines them in an intelligent way and segments customers into different client journey paths.

  • Prediction of demand for each product

    Given where the customer is at on their client journey, the model predicts the demand for various products, which serves as inputs to the recommendation engine.

Illustrative Optimal Trees for banking product recommendations

Recommendations that increased customer engagement

The client journey descriptions that accompany the recommendations provide the banker with more context on the fly and allow them to make more meaningful personal recommendations. The tool synthesizes the complex transaction history into something simple and relevant to both the client and the banker.

In a pilot study under the recommendations, the bank estimated a significant lift of customer retention rate and increased product uptake, especially in high-value products such as mortgage offerings.

Unique Advantage

Why is the Interpretable AI solution unique?

  • Actionable with validated insights

    The model outputs recommended products and rationale, providing the bankers with the context they need for better engagement

  • Signals from complex data

    Optimal Trees is able to sift through large volume and various sources of transaction data to learn what's most important in defining the client journey

Want to try Interpretable AI software?
We provide free academic licenses and evaluation licenses for commercial use.
We also offer consulting services to develop interpretable solutions to your key problems.

© 2020 Interpretable AI, LLC. All rights reserved.