Causal analysis on observational data for campaign planning in e-commerce
Understanding the relationship between cause and effect is one of the most fundamental human behaviors. This type of association, commonly referred to as causality, assists people in better understanding their environment and behaving accordingly. An understanding of causal relationships is beneficial in a variety of situations. With the ability to identify and interpret causal relationships with machine learning algorithms, all fields can increasingly rely on analytics results. Morgan et al. cite stock market analysts, pharmaceutical companies, advertisers, E-commerce, and other fields as suitable applications. Due to its complex applications and interdisciplinary issues, E-commerce is a discipline particularly suited for implementing and evaluating various causal machine learning methods.
Within E-commerce, numerous use cases exist for which causal analysis is suitable. Of particular interest are scenarios that examine the interaction between customers and businesses, which can lead to an increase in sales in the best case scenario. Such use cases are mainly found in price or sales planning. Discount-oriented campaigns are among the most extensive and profitable cases within these departments. Companies can use them to gain reach, find the sweet spot between discount and profit using optimal discounts, or reduce the inventory of legacy products.