How can lenders increase revenue through intelligent customer segmentation

Shamolie Oberoi   /    Content Marketing Specialist    /    2023-01-18


What are the key motivations of a business? It's a big question, and the answers are many, but they’re fairly common-sensical.

The primary goals of any business are to acquire new customers, retain existing customers, and of course, to make more money than they spend. And lenders - be it your new-age swanky NBFC or your traditional, legacy bank - are no different. They’re in a constant quest for new borrowers while simultaneously working on how to maximie profitability from the borrowers that are already engaged with them.

You can imagine it’s a mammoth task, especially considering the vast base of customers lenders work with - for reference, ICICI Bank has 52 million customers, HDFC Bank has 43 million , and Axis Bank has 20 million.

Given these vast numbers, it would be ideal if lenders could glean insights right at the outset to identify which of these customers would be the most profitable. More specifically, during onboarding itself, lenders should be able to identify which borrowers would be open to further cross-sell and up-sell based on current data patterns.

Clearly, doing this manually would be next to impossible- so, enter AI and ML-based customer segmentation.

What is customer segmentation?

Artificial Intelligence and Machine Learning algorithms can help slice a lender’s customer base into subgroups based on attributes such as digital engagement, conscientiousness, financial planning, behavioral attributes, and credit propensity, among others.

While this enables lenders to decide who to lend to and how much during risk segmentation-based underwriting - it can also play a much bigger role in business growth, helping lenders understand their incoming customer segments better and identify their needs at different points in time. 

Using probability of default on current loans and future cross-sell opportunity associated with a segment, the LTV of different segments can be estimated during onboarding which can then be used to categorise customers into desirable and undesirable segments.

The Idea is to make sure that any incoming borrower belonging to a desirable and high LTV segment is not lost to the competition. This can be done by making their onboarding experience seamless and personalised, from the very first interaction.


Let’s take FinBox DeviceConnect’s customer segmentation capabilities for example. The in-device risk engine securely analyses reams of customer data such as demographic indicators, app usage, bill payments, occupation type, and digital activity data to generate a comprehensive persona.

Based on these personas, lenders can decide who makes the most profitable customer (and that customer may be one with a slightly higher risk level - but also with a higher propensity to do business with that particular lender again).

Now, let’s take a deeper look at how comprehensive customer segmentation can benefit lenders:

  • Enables LTV-based underwriting

Current underwriting practices enable approval and offer decisions solely based on the probability of default or risk of a customer. 

However, if lenders can also estimate lifetime value through segmentation, tweaks can be made to the rule engine to acquire customers that will have a higher lifetime value to offset and surpass the higher risk cost associated with the segment

LTV-based segmentation can also enable LTV-based pricing, through which customers can be given attractive offers based both on risk and future cross-sell opportunities.

  • Reduces cost of acquisition

This happens in two distinct ways. Segmentation gives lenders a look into the defining features of the segments that are most profitable. Banks can then employ lookalike modeling to find potential customers with the same traits as your most profitable ones, who would be most likely to respond to targeted marketing messages.

  • Improves cross-sell and campaign strategy

Customer insights allow lenders to gauge what a borrower is looking for at a certain point in time, and align their marketing efforts accordingly. Such cross-selling and upselling efforts are key to growth especially considering that a  Gallup survey found that45% of consumers who were satisfied with their banking relationship also said they would consider their institution for their next product or service.

However, personalisation and timing are crucial to campaign success - studies have found that personalised upsell or cross-sell offers are 30% more effective when delivered within two seconds of a customer making an initial product or service selection.

Personalisation is key to ensuring the success of your cross-sell campaign. Even when selling the same product to different segments, it’s important to account for unique communication preferences - how would they like to be spoken to? Where should you communicate - WhatsApp or email? Borrowers are much more likely to respond to communication that’s customised to their needs. Segmentation defines these finer points for the customers solving a key problem for marketing campaigns.

  • Helps build relevant and next best product pipeline

Product managers are always on the lookout for new products or features that have synergies with their existing products. Segment analysis helps them identify features that appeal to different personas, so they can subsequently prioritise new products or features based on the desirability of that segment.

In-depth interviews and conjoint analysis for products can also be made more specific to the segment that is the most relevant audience for further monetisation These  customer insights combined with predictive analytics can help lenders make the ‘next best product’ for a particular segment.  

  • Optimises collection strategy

Segmentation is helpful beyond the initial stages of lending and marketing. It comes of immense use to lenders even at the collections stage. However, during collections, lenders may need to take alternate approaches to segmentation in order to build the most optimal strategy for debt recovery. These include:

Segment of one

Move away from a broad-based, static approach to segmentation that puts customers together in a black box based on past events. Instead, create ‘segments of one’, studying each borrower’s real-time financial behavior near the due date. ML models can help lenders glean insights to determine the probability of a default on a case-by-case basis (each borrower’s ability and intent to pay in real time).

Value-at-risk (VAR) segmentation

VAR - a measure of the potential and probability for losses to occur - helps lenders prioritise collections from high-risk, high-balance customers and refocus their communication strategies.

Behavioural segmentation

Lenders need to assess personal customer data and analytics to identify subjective experiences such as lack of liquidity or cash flow restraints that deter customers from timely repayment. They can create sub-groups within their cohort of at-risk borrowers and alter repayment plans or contact strategies on a single-bucket customer case.


Customer segmentation is about so much more than risk. It is the  key to overall business growth. A comprehensive segmentation solution can help lenders offer borrowers customised products, maximise cross-sell and upsell, share the right promotional offers on the right channel, and also identify new, profitable segments.  At the end of the day, treating every customer equally doesn’t make good business sense - personalisation is key to success.  To learn more about how DeviceConnect can help supercharge revenues through intelligent customer segmentation , get in touch with us  here.