BKB - Analytical CRM

Challenge
When it comes to marketing, banks are like any other business - they need to be in constant contact with their customers to be able to respond to their needs. They need to know what their customers' lifestyles and habits are. The banks' marketing campaigns should also be geared towards this in order to offer customers products and services that are suitable for their current life situation. The Basler Kantonalbank (BKB) is also striving to implement personalised and targeted campaigns. Pascal Freudiger, Data Scientist at BKB, explains: "In the past, we selected customers based on general attributes such as gender or age group. And because we created the data manually, it took a lot of time to prepare the campaign. This was not always efficient." To counteract this and intensify marketing activities, a new plan was needed. After evaluating several solutions, BKB chose IBM® Customer Insight for Banking. BKB found this to be user-friendly, efficient and easy to maintain. It was implemented by iRIX Software Engineering AG.


Solution
Jörg Koch, an expert in analytics solutions for the financial industry at iRIX, explains the approach: "We participated in a proof of concept with predictive analytics tools to help BKB select their customers for specific campaigns." In one A/B test, customers were selected using the traditional method, while the other group used leads generated by the IBM Customer Insight models. Pascal Freudiger confirms: "With the traditional way we recorded a success rate of 7 per cent - with the Customer Insight models we were able to achieve a success rate of 35 per cent."

Instead of manually selecting customers as was done before, BKB can now pull data on its customer behaviour from the central data warehouse to create a master data record of 650 fields. These fields can be selected by account value or turnover, for example. Based on this descriptive data, specific statistical models are created for certain use cases. These use cases can then be used to optimise marketing campaigns. For example, these models tell us how customers react to advertising or how product-savvy they are. Finally, customer advertising can be specifically tailored to them. Pascal Freudiger explains: "We use the models again and again to evaluate and classify customers based on more than 600 key attributes, such as customer master data, business volume or aggregated transaction volume."bewerten und zu klassifizieren.“


Added value for the client
Thanks to the use of predictive models, the bank achieved a response rate that was 5 times higher within the "proof of concept" than within the manually selected control group. Since then, machine learning methods have been successfully used in campaigns and sales promotions. The analysis services are also used by sales for spontaneous sales campaigns. The manual selection of customers had previously taken an enormous amount of time. This is no longer necessary thanks to the new solution and the effort required was thus significantly reduced. In addition, the sales department achieved efficiency gains by improving their sales processes with behavioural analyses and score values of the customers. As a result, customer relationships can be better tracked and at the same time repeated contact can be avoided. Pascal Freudiger is pleased with the result: "Although we now contact fewer customers in a single campaign, the turnover per campaign has not decreased significantly because the campaigns are now more targeted and efficient. It's quality over quantity."