Assisting Ad Testing with Viewer Emotional Response Prediction: A Guideline and Method Development
(A2023-114229)

AUTHORS

Yuanyuan Zhu, The University of Western Australia; Paul Harrigan, The University of Western Australia; Kristof Coussement, IESEG School of Management, Univ. Lille, CNRS, UMR 9221 - LEM - Lille Economie Management; Tina Tessitore, IESEG School of Management

ABSTRACT

Testing ads before they are finished or released is a significant part of the market research industry. It is important to test emotional response to provide preliminary directions as it plays a fundamental role in the chain of advertising effects. However, there remains a pervasive underutilization of ad testing, especially in the area of testing emotional responses. There are two main challenges. First, it is not clear to practitioners which specific emotional appeals are most suitable for use in different contexts. Second, it has always been a challenge to measure emotional response. Therefore, this paper aims to address these challenges by 1) providing an overview of the current state of knowledge regarding the use of emotion appeals in advertising through a systematic literature review, and 2) building and validating a machine learning model for emotional response prediction to assist firms with ad selection. We contribute to computational advertising research by introducing machine learning and image mining techniques into ad testing. Practically, we provide a guideline and a method to assist ad testing with viewer emotional response prediction.