HSBC is a global banking and financial services organization with 7,500 offices in over 80 countries. The bank’s businesses range from personal and private banking to corporate and investment banking.
HSBC wanted to develop a Cross Sell strategy to sell its liability products to an asset customer base. More specifically HSBC was seeking to cross sell its Power Vantage Account (PVA), a Savings type account, to its Credit Card base. PVA is a premium product that offers many advantages aimed at the more affluent customer.
Xerago was brought on board to undertake a detailed study of the challenge. Xerago proposed building a response model across the Credit Card customer base. The aim of this model was to identify Credit Card customers who were most likely to respond to a PVA campaign.
The base customer data (all credit card accounts with 6 month vintage) was identified as 1.177 million. Xerago eliminated all the duplicate accounts and identified the absolute number of credit card accounts (with 6 month vintage) dating back to two years. A month on month distribution of responders and non-responders were also identified. Finally the responders (those who had accepted the PVA offer) accounts were observed over a 6 month window.
With the above analysis Xerago identified that the response rate for the entire customer base was just 0.19 %. Thus Xerago decided to further segment and spruce up the data to derive more meaningful insights.
Segmentation and Clustering
Xerago undertook a clustering approach to ensure the statistical representativeness of the population. However due to large amount of dormant customer and minimum balance owners, the cluster data was found to be skewed. The PVA offering being a premium product was deemed non compatible with this portion of the population and hence were dropped from the overall base. Xerago strove to ensure a healthy population base and not lose too many responders. The final base numbers were
Customer base: 654010
The pruned base was further subjected to a 30% sampling rate for final segmentation.
Factor Analysis and Final Clustering:
Xerago derived a basic set of 9 variables for use in the factor analysis and in further clustering. After working with the bank Xerago was able to drop out 3 of the variables based on the internal criteria without compromising on the cluster robustness. Xerago was able to derive good differentiating segments which were subject to further sampling.
From each segment, 30% of the non-responders were sampled. To boost response rates, the number of responders was kept unchanged.
The response model building was an iterative process. Different sets of input variables were used to derive combinations of new variables before arriving at a final set of input variables. The identified variables were chosen with the objective of cross selling across the card base. Hence most of the variables chosen pointed to active and moneyed customers, who were the original target group. The input variables were also subjected to basic Multi-Collinearity tests to ensure the variables’ invulnerability.
This resulted in creating the final model which was then tested across the rest of the base that was not part of the modeling base. Once the robustness and accuracy of the model was tested, it was then used to score the new base.
Thus, Xerago delivered a model that was the basis for the liability cross sell strategy.
The bank was able to use Xerago’s model to undertake an analytical and streamlined approach to targeted cross selling and power cross sell campaigns.