Abstract
Recent advances in generative AI have enabled marketers to produce content at unprecedented speed. However, conventional content testing methods such as A/B testing and multi-armed bandit testing have not evolved to match this accelerated pace. This dissertation addresses this challenge by proposing two novel AI-driven systems designed to enhance the speed and efficiency of content testing in marketing. In the first essay, I propose the Email Open Rate Predictor (EMOP), an AI system that leverages recipients' characteristics, headlines’ topics, emotions, and campaign sending times to predict email campaign open rates. EMOP enables marketers to instantly evaluate countless variations of email campaigns without needing large samples or waiting hours for results. The second essay presents DeepAudio, an AI system specifically developed to assess the likeability of audio ads, filling a critical gap since prominent online audio platforms currently lack built-in ad testing capabilities. DeepAudio integrates established psychological insights on ad effectiveness with advanced AI algorithms to automatically evaluate multiple audio ads. Extensive validation confirms DeepAudio's robustness and generalizability, providing marketers with immediate and actionable insights. Together, EMOP and DeepAudio provide marketers with efficient and effective tools, significantly enhancing their ability to optimize content performance at the speed of modern AI content creation.