A novel data-driven approach to detect and predict customer transitions in the marketing funnel

Abstract This study braids together insights from two distinct areas of research: customer engagement within brands and social data analytics. Leveraging a proprietary B2B dataset from LinkedIn comprising over 200 million advertiser interactions over a subset of de-identified members, we present a data-driven exploration of the dynamics of the so-called marketing funnel. By bridging the gap between real-world digital metrics and theoretical marketing funnel models, we offer empirical insights into the interplay between engagement types and customer transitions towards purchase. Our analysis delineates the structural underpinnings of the derived marketing funnel to shed light on the efficacy of various engagement strategies in propelling users towards deeper funnel stages, aiding advertisers in predicting and targeting engagements effectively. We make three contributions: first, we propose a novel data-driven methodology to create a representation of the marketing funnel. Second, we apply the data-driven marketing funnel model on a proprietary dataset from LinkedIn advertisements with a subsequent descriptive and predictive analysis of the funnel stages and transitions to showcase the practicability and usefulness of the approach. Third, we analyze the relationship between user engagement and stage transitions within the created marketing funnel, identifying what user engagements are predictive of future stage transitions.