The journey towards discovering a novel marketing attribution method.

Abstract

The global advertising industry is projected to reach $2.5 trillion by 2030. It reflects the growing dependence of businesses on media and advertising platforms to attract and retain customers. The rise of advertising as a dominant business model, particularly among large digital platforms, has significantly increased the cost of market access. As a result, businesses left with little choice but to invest across a growing number of marketing channels. This challenge is particularly prominent for digitalnative businesses seeking broad audiences in competitive markets. Despite advances in research and technology, smaller-scale advertisers remain poorly equipped to answer one of the most fundamental questions in marketing: what is the best use of our advertising budget? This thesis addresses this gap through three empirical studies. The first investigates whether web traffic generated by various marketing channels influences direct traffic and purchase behaviour, establishing the interdependencies that exist across channels. The second examines whether digital media investments generate spillover effects on organic branded searches, revealing how paid activity shapes unpaid performance. The third, driven by the findings of the first two, proposes a causal-driven attribution framework that estimates channel influence without relying on user-level data, offering a privacy-compliant and accessible measurement approach for businesses of all sizes. Together, these studies contribute to a more rigorous and equitable understanding of multi-channel marketing measurement. This work contributes to both marketing practice and academic literature. For practitioners, it offers actionable frameworks to measure channel performance, allocate budgets more effectively, and make evidence-based decisions without requiring technical expertise or user-level data. For researchers, it advances the understanding of multi-channel attribution, media spillover effects, and causal inference in advertising measurement, areas where empirical evidence remains limited.