Forecasting Chargeback Volumes: A Predictive Analytics Framework

Priya is a Domain Consultant for credit card operations. She has been with TCS for over 12 years, managing operations in the retail banking domain across merchant operations, loans servicing, cards transaction processing, records management, customer service, and disputes. She is an engineer and has a Master's degree in Business Management. Effective and timely planning and superior execution determine success in banking. Hence, volume forecasting and capacity planning are critical to the industry. Changing customer spending-patterns and heightening fraud risks have made it even more imperative for banks to relook at their forecasting models. Banks use accurate forecasting as an important parameter in capacity planning and pricing for prospective customers. Capacity estimates are equally important for the banks to monitor and compare team performance across processes and drive improvements. Depending on the data behavior and business inputs, the bank can choose an appropriate forecasting methodology to improve the accuracy of forecasts. For complex transactions such as chargeback processing, traditional methodologies such as time series may not provide accurate forecasts for chargeback volumes. This paper introduces a forecasting model that predicts chargeback volumes effectively and is aligned with the operations of a banking unit. The proposed model is simple but meets business requirements, requires minimal investment of resources, and increases the accuracy rate. It also addresses the entire chargeback process. The model adopts a blended approach based on historical patterns (within stages of the process cycle), predictive analytics, and market trends. This alignment with business patterns helps in monitoring process performance and estimating the price per transaction for new engagements at the bidding stage. It also helps businesses estimate the required capacity or staffing levels. Abstract Contents