The main purpose behind this study was to analyze the problems faced by big retail banks during business expansion. Typically banks tend to acquire new customers at huge costs rather than leveraging their existing customer base. A bank’s customers leave behind a large footprint in terms of the transactions they perform, which can be analyzed to understand their behavior pattern which may be leveraged for selling new products.
This paper analyzes the customers’ transaction patterns, product holdings, demographics, past trends, and other attributes to devise an effective strategy for engaging them further. In a retail bank with various product offerings, the focus should be on customer segmentation and profiling to ensure ease of targeting, marketing, and offering personalized products to retain profitable customers and capturing market share across geographies.
Keywords: customer segmentation, profiling, clustering, business expansion, profitable customers, scorecard, bank, customer, transactions
Table of Contents
- Literature Review
- Profiling To Define Profitable Customers
- Technique Used: Scorecard
- Scorecard Analysis Summary
- Profiling Profitable Customers Into Various
Segments To Customize Product Offerings
- Technique Used: K-Means Clustering Algorithm
- Clustering Output and Summary
- Application And Conclusion
- Profiling to classify each customer into either profitable or non-profitable buckets
- Profiling profitable customers into various segments to customize product offerings to increase the overall business of the bank
Benefits of customer profiling and segmentation:
- More customer retention
- Enhances competitiveness
- Establishes brand identity
- Better customer relationship
- Leads to price optimization
- Best economies to sale
- Improves channel of distribution
- Increase profit by keeping costs down
- Identify potential customers
- Improves Customer Engagement and Brand Loyalty
PROFILING TO DEFINE PROFITABLE CUSTOMERS
To determine customer profile through a Scorecard, we have covered the below data points:
- loan account with bank
- loan duration
- loan amount given to customer
- whether customer is having credit card
- whether customer is having multiple accounts
- credit quality of customer
- whether customer created any standing order with bank
- relationship tenure with bank
- average monthly balance in customer account
The following rule would help to understand the scoring methodology for continuous and categorical variables.
Continuous variables: Box plot statistics have been used to score them between 1 to 4. If the customer falls into the 1st quartile, the customer has scored 1, for quartile 2 score would be 2 and for quartile 3 and 4 scores would be 3 and 4 respectively.
Categorical variables: The answer is either Yes or No. If the customer is having a credit card, that customer would be given 1 else 0.
After providing a score to each customer against each variable, we have then assigned the weightages to each variable between -3 to +3. Following are details on the rationale for the weightages attained, against each criterion:
As per the total score bracket (7 to 36), we have divided them into 4 quartiles. Q4 has scored the highest and Q1 is having the least total score bracket. Q2 and Q3 remain in the middle portion. If a customer’s is falling in Q4 bracket, then the customer would be a profitable one. This would help the bank in the segregation of its base in various segments as per the profitability generated by client relationships.
From the scorecard analysis results, out of the base of 4500 customers, we could identify 3184 customers (falling in Q2, Q3, and Q4) are the potential and profitable customers who add value to the bank’s profit.
- The bank should target these customers with customized offerings to further increase its revenue.
- The bank should further weed out the base of non-profitable 1316 customers (falling in Q1) to reduce the cost that incurs in their retention.
- The bank should further segment the profitable customers to move them up to the higher profitability bands, for example from Q2 to Q3 and from Q3 to Q4 by suitably nurturing them.
PROFILING PROFITABLE CUSTOMERS INTO VARIOUS SEGMENTS TO CUSTOMIZE PRODUCT OFFERINGS
Out of the base of 3184 Profitable customers, Cluster 1 is having the highest population of customer concentration with a total number of 2199 customers and on the other side Cluster 2 is having the least number of customers with a total of 462 customers. Cluster 3 is having 523 customers.
We have profiled these clusters descriptively:
APPLICATION AND CONCLUSION
These analytical models or techniques (scorecard and clustering) used in this study are generic in nature and not specific to the case in point. These models can find a wide application across the financial services industry.
You can also read this article on our Mobile APP