Indian Banks today offer a gamut of financial products customized to the lifecycle and lifestyle needs of their retail customers. Credit card is one such product that is targeted towards specific customer needs, namely travel, lifestyle, fuel etc.

Banks earn commission each time credit card is swiped and hence try their best to keep their customers active by luring them with offers and discount across partner merchants. A digitally savvy customer today has a plethora of options when it comes to choosing a credit card with most offering a similar set of features. Having two or more credit cards isn’t very uncommon but would you as a bank like losing or sharing your portfolio of spend with other players. What else can a bank do to retain and grow wallet share of credit card spend when there is little to differentiate in credit card offering.

Do you think serving contextual & relevant merchant offers to your customers can keep or grow your wallet share? Most of you will agree with me on this. Personalization or recommendation engines is a promising solution most of us have already heard of or perhaps tried out as well. Our survey of some banks who have used personalization algorithms indicates results are far from what is being promised. We tend to blame these algorithms but what is really wrong? Algorithms are state of the art. Our processes and IT systems have improved too as well.

Having crafted such engines personally, I can confidently say that data quality is the single most important reason for low returns from such technology investments.

Consider this

  1. Merchant Name with each transaction is a pretty important field that helps me identify customer’s preference for an entity in a category. However, what can you do when we have several versions of one single name. Example Flipkart, www.flipkart.com, Flipkart private limited. All are same but presented differently. You cannot even think about segmenting your customers by merchant name.
  2. Merchant Category is another important field useful for personalization. Are your merchant category names across debit and credit card standardized? We haven’t found so in our case. Some examples: Fast food restaurant vs Dining & Bar, Hospitals vs Healthcare, Passenger Railways vs Railways, Hotels & Resorts vs Hotels. If your customer spends using both credit and debit cards across merchants, you are often not in a position to identify his/her preferences clearly.

Clearly garbage in garbage out! Your team needs to invest time to clean up the data to be fed into algorithms.

Let’s say we have tackled the data quality problem. Where does that leave us? How about data richness? Do we have enough information about customer’s preferences to be able to personalize merchant offers? Clearly no. Example if a customer has spent 5000 Rs. at Shopper Stop or Flipkart we never will know what he purchased. If I have a set of offers on apparel I will never be in a position to personalize these at the finer level (Remember I don’t know brand, color or size preferences). Nothing to crib about something we can’t get.

Poorer the data, poorer the outcome!

Can we enrich our data? Yes

  1. We do have certain merchants who are known for single product category like Cafe Coffee Day, Spice Mobile, Dominos, Bookmyshow.com etc. We can surely enrich these merchants to allow us to create personas like pizza lover, coffee lover, movie lover. In order to do this, our team will have to manually associate merchants with relevant tags. Business team can work closely with analytics team to define such personas.
  2. Third party data sources. There can be many here if we talk about serving contextual and relevant offers. Weather data, merchant data, traffic data can all help us improve our targeting. If we have offers from different merchant’s in dining category we can use ratings and reviews from zomato coupled with customer’s location to recommend personalized dining offer.

All this is easier said than done. It will need sufficient time and investment to improve data quality and enrich your data. Rest assured if done correctly you will see tremendous improvement in whatever personalization platforms you use. Hope you find this useful and if you have any ideas you would like to share with our readers feel free to share.

JOIN OUR COMMUNITY
I agree to have my personal information transfered to MailChimp ( more information )
Join over 3.000 like minded AI enthusiasts who are receiving our weekly newsletters talking about the latest development in AI, Machine Learning and other Automation Technologies
We hate spam. Your email address will not be sold or shared with anyone else.
Tags:

Leave a Reply