Enhanced Last-Touch Interaction Attribution Model in Online Advertising

摘要

The increased popularity of an internet opened a new way for e-business in terms of digital advertisement. In order to get back and improve the “return of investment”, how to allocate the revenue distribution to different marketing channels comes out to be the key problem in digital advertising. However, last interaction model, first interaction model, last click, last ad words’ click, linear attribution, time-decay attribution, position based attribution models are some of the attribution models developed to attribute and assign contribution to each marketing channel. These existing models consider the contributions of the other channels and some don’t consider the synergistic effects in revenue calculation from different marketing channels. This paper proposes Enhanced Last Touch Interaction (ELTI) model to allocate the revenue distribution to different marketing channels using game theory and synergistic effects. Additionally, the paper also implements and adopts the probabilistic approaches to prevent the simple intuitions made by many other attribution models. Prediction accuracy of above 75% of the ELTI model out performance the state-of-the art models.