A Comparative Study of Data Analytics Techniques for e-Marketing Optimization

Abstract

Data-driven decision-making in the e-marketing landscape is becoming increasingly important. Businesses are able to collect a variety of data from their users across various platforms, including their website and even on social media. They use this data to make more informed decisions when designing e-marketing strategies and practices, and also to optimize the current strategy in terms of performance (i.e., ensuring that the messages are reaching the right people at the right time). The data sets collected over time grow rapidly in both size and cardinality, and one possible path to a non-technical person's understanding of the data is through descriptive analytics and predictive analytics techniques. Our focus is the comparative study of such data analytics techniques for e-marketing strategies that engage in e-CRM thinking and focus on systems already in use or existing consumer user bases, i.e., the application of models within a context of "operations marketing management."

In this work, we provide an overview of the e-marketing optimization initiatives that will aid the reader in contextualizing the processed data. Following that, we present a comparative study of the three main data analytics techniques (descriptive/prescriptive data analytics, machine learning, and time series methods) to predict new customer visits to a computer manufacturer's website based on collected consumer data. Our results are promising, demonstrating that, when carried out correctly, the predictions from both descriptive and predictive analytics can contribute up to a 30% incremental trend over marketing campaign efforts for one week on a particular website.