The Customer Journey as a Source of Information

In the face of heightened data privacy concerns and diminishing third-party data access, firms are placing increased emphasis on first-party data (1PD) for marketing decisions. However, in environments with infrequent purchases, reliance on past purchases 1PD can pose a challenge in fully harnessing this data. We address these challenges by introducing a probabilistic machine learning model that fuses customer click-stream data, and, when available, purchase data, within and across journeys. Combining data across journeys can be challenging if customers have different needs in different journeys. We account for such “context heterogeneity”, using a Bayesian non-parametric Pitman-Yor process. Drawing from within-journey, past journeys, and cross-customer behaviors, our model offers a solution to the “cold start problem,” enabling firms to predict customer preferences even in the absence of prior interactions. Notably, the model allows for continuous updating of the inferred preferences as the customer keeps interacting with the firm. We apply the model to data from an online travel platform, revealing the substantial benefits of consolidating 1PD from both current and previous customer journeys. In scenarios with restricted access to customers’ historical data collected via cookies, our model suggests that within-journey data can help offset the loss of historical customer data.