Blog article
See all stories »

AI powers proactive hyper-personalized experience for retail banking customers

A recent satisfaction study by J.D. Power for U.S. retail banks found that banks have struggled to deliver on customer expectations for personalization and almost half of the customers have moved to digital-centric banking relationships. Today, the expectations of banking customers have changed, where they’re now looking for hyper-personalized offers like those provided by Netflix, Amazon, and Starbucks. Hyper-personalization can be delivered by harnessing artificial intelligence (AI) and machine learning (ML) with real-time data and tailoring customer experiences. This blog explores the opportunities in leveraging ML models to hyper-personalize customer experience across customer channels, namely, contact center, web, and social media.

Shift in customer experience approach

Customers expect a meaningful and highly personalized digital experience for their individual banking needs. Banks can predict these needs by understanding their customers better- their goals, preferences and behaviors in real time and proactively delivering tailored offerings. Consider a scenario where a customer is spending more money than usual which could lead to them having insufficient funds for their upcoming EMI. What if the bank can predict the expenses based on the past spending trend. The bank can then proactively alert the customer and offer discounts on a personal loan. Such a proactive, contextual, and personalized experience initiated by the bank can deepen customer relationships.

Considering this has been a topic of interest in the recent past, let’s explore how AI/ML research is applied to three different customer channels independently and then compare the three approaches.

AI-based hyper-personalization or recommendation models

1. Customer services call center: Predicting the reason for a customer call and performing pre-emptive intervention would entice customers. Researchers have developed an AI-based multi-task Neural Network (ANN) to predict a customer’s call’s intent and subsequently migrate the customer to digital channels. The machine learning model was trained using the customer’s profile, call transcript data, customer servicing log and transaction log. The objective is to predict if the customer will call the contact center in the immediate future, say within the next 10 days.

When the customer calls the IVR system, a personalized voice prompt will recommend relevant digital services based on the model’s prediction. If the customer accepts the recommendation, then they are redirected to launch a chatbot through an SMS with a URL. This results in hyper-personalized and efficient customer service experience. Consider a scenario when a customer has deposited a check but the amount hasn’t been credited to their bank account even after a week. The customer would enquire by calling the contact center. The machine learning model would predict the call’s intent for this specific customer and move to their preferred digital channel for an appropriate resolution.

2. Web channel: Personalization based on user behavior is generally done using data mining algorithms, but user behavior prediction for full personalization is very difficult. This is due to frequently changing usage data with changing user interest. Researchers have found a novel intelligent web personalization model for user preference recommendation. The machine learning model predicts the web content for the user and learns the user behavior continually. Banks can use the model to recommend products tailored to a specific user.

Instead of offering personal loans to every customer who enters their website, the banks can personalize the home page for their customers based on the browsing history and their current stage of life. For example, a customer with a young family would be more interested in taking out a mortgage or car loan or long-term investments. A customer who is retiring soon may require help with retirement and wealth management plans. Using the above AI model, banks can tailor the website dynamically by recognizing the customer and anticipating the need.

3. Social media channels: These platforms generate a wealth of customer related data including behavioral data which can be used by banks to gain a deeper understanding of customers’ needs. These valuable insights can lead to proactive personalized offerings for customers. Researchers have developed an integrated framework to help banks in deriving value from social media analytics. This will help to tap into advanced AI-based prescriptive and predictive analytics to develop insights for hyper-personalizing customer experience. Consider an example of a customer posting comments on Facebook about specific tourist destinations and their interest in visiting these places. This is a great opportunity for the bank to analyze the posts and suggest tailored offerings like personal loans, travel insurance and offers on travel tickets.   

In these three customer channels, the data required for the predictions varies from one channel to another. Figure 1 gives the summary of the data involved in customer engagement on each channel. We see that there’s higher data complexity in contact center and social media channels because of unstructured data.

Enrich customer experiences: The way forward

We discussed the machine learning models recommended for different customer channels. As the data sets, data types and user behavior in each channel are different, every customer engagement is unique. We see increasing complexity in AI models as we move from web channels to contact center channels to social media channels. The banks can consider these while prioritizing and deploying machine learning models for hyper personalization.

AI based prediction models using real time data look very promising. It provides an opportunity for banks to tailor every customer touchpoint. We deliberated on hyper personalization across the three channels and the enormous value which can be unlocked. This can enable banks to hyper personalize, improve customer stickiness resulting in significant growth.

a member-uploaded image

Comments: (1)

Ketharaman Swaminathan
Ketharaman Swaminathan - GTM360 Marketing Solutions - Pune 29 November, 2022, 10:32Be the first to give this comment the thumbs up 0 likes

Have you heard of Overdraft Protection Fees and Cardlytics?

Overdraft Protection Fees would explain why a bank will NOT want to offer a discount on personal loan.

Cardlytics would explain how 2000 US banks are already making Targeted Offers based on transaction history.

Senthil C

Senthil C

Senior Consultant


Member since

18 Jan 2022



Blog posts




This post is from a series of posts in the group:

Banking Strategy, Digital and Transformation

Latest thinking in respect to Banking Strategy, Digital and Transformation. Harnessing our collective wisdom to make banking better. Ambrish Parmar

See all

Now hiring