A Causal Framework for Digital Attribution

Jim Koehler
Google LLC (2018)

Abstract

In this paper we tackle the marketing problem of assigning credit for a successful outcome to events that occur prior to the success, otherwise known as the attribution problem. In the world of digital advertising, attribution is widely used to formulate and evaluate marketing but often without a clear specification of the measurement objective and the decision-making needs. We formalize the problem of attribution under a causal framework, note its shortcomings, and suggest an attribution algorithm evaluated via simulation.