While a great deal of focus currently rests of predictive analytics and machine learning, the resulting models are usually only part of a solution. Knowing when to utilize a particular model, and in which circumstances is also of crucial importance. The case studies shown below, courtesy of FICO the sponsor of this report, demonstrate the key role played by optimization in finding real-world integrated analytic solutions:
A leading US retailer wanted to leverage its extensive retail loyalty club program to improve the take-up and use of product offers. Its vision was to align its whole organization around a more member-centered approach. as part of its FICO Analytic Offer Manager solution, FICO generated thousands of propensity models at the product sub-category level for every offer available, which focus not just on what prospects are most likely to buy, but when they are most likely to make the purchase. The outputs from these are then used within FICO Xpress Optimization Suite to match offers to consumers in a manner that maximizes key performance indicators (KPIs) — such as maximize incremental sales that have a positive margin, whilst allowing for many member- and business-focused constraints.
The solution has provided positive results for the retailer across multiple campaign concepts and objectives, increasing response rates by factors of more than x5 on many segments. It has shown significant on-going benefit – Across different customer segments. – For both low and high tickets items. – And for new product trials.
A large European Bank aspired to implement a more customer-centric marketing campaign approach and needed an ultra-large-scale optimization solution to meet their business requirements. A decomposition technique, and use of the parallelization features of the FICO Xpress Optimization Suite, was used to solve the optimization model. This confirmed that the solution could run within the current infrastructure, on the whole customer base, within the required timeline. Additionally, the Xpress-Insight module was used to quickly develop a series of scenario planning and comparison dashboards. The results of the proof-of-concept led to the full implementation of an integrated FICO Xpress Optimization Suite solution.
A bank in the Asia-Pacific region wanted to Maximize profit on authorization & Credit Limit management decisions balancing between:
- Controlling accumulation of bad debt
- Attrition/customer service
- Revenue Generation
A decision optimization solution meant an additional $500K approved transactions in the first month, a 7% increase in income over five months with no increase in delinquency, and a projected additional $110M in transaction values over 12 months, with similar approval rates. A project for credit line increase realized .$1.5MM in incremental profit in the first year, and the strategy was rolled out to 100% of the portfolio. Both projects created decision tree strategies for direct import into FICO TRIAD Customer Manager.
These examples illustrate well the advantages of combining various analytics techniques, something that needs an integrated analytic platform – in this case FICO.