The institution currently does not have customer segmentation in terms of their marketing and targeting processes. Thus, all their marketing efforts are generic and not cost effective to the organization. In addition, they do not have a standardized framework to understand the long-term value (LTV) their customers generate for the organization. If known, that would help them understand the various customer segmentation based on their LTV values, which would help them target consumers effectively and thus increase their profit margin. They needed a proper statistical framework to automate their marketing and targeting processes for better ROI.
The solution consisted of four phases. First phase included preparing and transforming the data. The data retrieved from Oracle database needed polishing and few transformations were required to ensure a better prediction. The second phase included feature selection. Not all attributes might be useful, so feature selection helped in identifying important attributes that will help in model building. The third phase was the most important and crucial phase where we needed to form specific clusters of data that are homogenous and separable. This phase helped us in segmenting various customers of the bank. The last phase will calculate the specific LTV’s of the segment through a mathematical construct viz. Statistical and Machine Learning algorithms.