Browse Prior Art Database

Pricing Optimization for Banks Using Customer Behavioral Insights

IP.com Disclosure Number: IPCOM000246588D
Publication Date: 2016-Jun-20
Document File: 5 page(s) / 432K

Publishing Venue

The IP.com Prior Art Database

Abstract

Disclosed is a system to analyze customer data to generate price optimization options for retail banking.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 52% of the total text.

Page 01 of 5

Pricing Optimization for Banks Using Customer Behavioral Insights

Globalization is a challenge for the retail banking business model because it is changing customer expectations, opening competition from non-banks, etc. Banks need a new way to optimize pricing to maximize revenue.

State-of-the-art pricing technologies use dynamic, relationship based pricing strategies, but none of the current solutions uses deep customer analytics and insight at the individual customer level.

The disclosed solution presents an innovative approach to price optimization for retail banking.

The novel system analyzes customer data (i.e., transactions, interactions) to create product/service level churn propensity and buy propensity models. The system analyzes account origination, transactions, fees, and charges data to create a Bayesian classifier that predicts whether a customer will accept a specific pricing offer. The system then uses Monte Carlo (MC) simulation to create the probability distribution of the client's accepting a given offer. Finally, the system uses this distribution to make a pricing decision. This solution can be operationalized at the point of service/sales (POS) level in real time.

Task 1: Churn Propensity Prediction

Figure 1: Step 1: Tune the churn propensity model

Figure 2: Step 2: Apply the model to segment existing customers based on propensity

to churn

Figure 3: Step 3: Apply the model on an ongoing basis to new events/customers to track changes in propensity. This step is frequently repeated, while steps 1 and 2 are mainly
one-time operations, but can be repeated as part of the learning.

1


Page 02 of 5

Task 2: Product Propensity Prediction

Figure 4: Step 1: Tune the product propensity model

Figure 5: Step 2: Apply the model on existing customers to create an associated

product propensity score

Figure 6: Step 3: Apply the model on an ongoing basis on new events/customers to track changes in propensity. This step is frequently...