Engagement overview
The bank’s existing loyalty and rewards programme has an unsatisfactory penetration rate and is not leveraging data sufficiently to entrench customers. The bank also has an aggressive drive for improving market share through increased primary banked customers.
On the back of earlier work done for the bank, BSG was asked to assist by leveraging customer intelligence and data, to enable the bank to better understand their customer’s financial behaviours and spending patterns.
To drive business value, an organisation requires data preparation and enrichment as the cornerstone activities to consolidate and transform disparate data into a format that will drive business value. For this project BSG developed analytics models to segment customers based on their value, behaviour and affinities. Insights were then drawn, using these models and customer data, to answer key business questions.
The customer insights revealed would not only be of value to loyalty and rewards, but given the extended nature of data collected, the whole bank would be able to benefit from a deeper understanding of their customers.
This was a comprehensive data project, with BSG collecting over 6 million individual retail customer records. These customers held at least one active account during the preceding 12 months. 46 data tables were cleaned and consolidated into a single customer data set with over 200 attributes per customer. Further activities included dimensionality reduction and optimising size wherever possible.
A significant amount of time was spent on feature engineering to enrich the data by creating behavioural variables that were not obtained directly in the transactional data. These were derived and used to indicate customer financial behaviour.
Throughout the data collection and feature engineering processes a data dictionary was composed detailing a list of all variables and calculations used to enrich the final dataset for statistical modelling. The data dictionary was created to serve both the technical and business users and ensure reproducibility of the project and the ability to orchestrate future business questions.
Supervised segmentation models were built and nine key financial behavioural segments were identified and profiled to understand saving, spending, borrowing and product interaction dimensions.
Subsequently, a self-service analytics model was built to enable key business questions to be interactively answered. Significant value was produced by providing a single view of the customer, spanning across all the main product houses of the bank as well as demographical information, with t he ability to drill down to an individual client level.
This project has created the foundation to support ongoing data analysis to assist in answering existing and future business questions from multiple angles.
Making a difference
What was unique about this project was BSG’s approach of innovative thought processes to push the boundaries of how to. Instead of merely relying on the information given to BSG, innovation was utilised to creatively change the given data structure to release hidden information through transformations and feature engineering. This was all achieved through continuous engagement and collaboration with the bank’s key stakeholders.
By expanding the BSG toolset and using what is most appropriate , the information was extracted effectively and efficiently to present the information in a story that non-technical business owners could relate to.
Focus was maintained on surfacing only information that is actionable, implementable and non-trivial i.e. real intelligence that the bank can utilise to better understand their customers in a way that is beneficial both to the bank and the customer.