Evolution of Customer Segmentation in the Era of Big Data

Abstract

The development of information technology (IT) causes an increase in the amount of data to be created, stored and processed for the needs of various organizations. Segmentation is one of the marketing tools can help the organization to promote sales activities and benefit from it. It is important for marketing practitioners and decision makers to understand the concept of predictive modelling and have an understanding of how to use big data for segmentation purposes. Marketing and Information Technology are blending due to digitalization, statistics is becoming more important due to the rise of big data and data mining opportunities. Boarders of different disciplines are becoming vaguer and interconnection of disciplines can be observed more often. The purpose of the study is to create customer segments based on predictive modelling by using big data available in an organization. Data for modelling is used from a non-banking lending company based in Latvia AS 4finance. The process of data mining is described and performed in the study using data provided by the company. For the data mining process and the development of customer segments, the authors selected RapidMiner Studio software and used CRISP-DM data mining methodology. Three types of activities were tested to evaluate the economic benefit of created segmentation model on overall 11321 customers. All customers were segmented into two groups based on the created predictive model-one group contained customers that were predicted to become an inactive and second group with customers that were not predicted to become inactive. All customers were split into three groups containing a similar split of predicted outcome. Three different types of activities were performed with all three groups. As a result, common characteristics of segmentation and predictive modelling were identified. The results of the empirical study show that it is possible to create customer segments by using sophisticated predictive model. This can be achieved without having to write statistical software codes. The study results also show that the organization can benefit from the implementation of segmentation based on data mining and predictive modelling in key business areas. Segmentation model created during research show economic benefit for the company. Authors also indicate that this segmentation approach can be replicated in different business areas.