Marketing Experimentation and Machine Learning: A Deep Dive Into Data-driven Marketing Within the Independent Publishing Industry

Megan Holly Jackson

Studenteropgave: Kandidatafhandlinger

Abstract

A/B/n experimentation has long been considered a golden standard for display advertising optimization. Recent literature in this area, however, has demonstrated a noticeable shift. Discourse concerning this experimentation technique has grown less optimistic, with many emphasizing the inherent limitations that the analytical framework underpinning this technique exhibits. Such studies are now calling for the inclusion or replacement of this methodology with advanced machine learning. The following thesis seeks to contribute to this body of literature by conducting an exploratory study that endeavors to answer the research question: To what extent can advanced machine learning be employed to optimize A/B/n testing of display advertising? To facilitate answering the research question a theoretical framework is created, conceptualized as machine-facilitated experimentation. Key concepts here are used to guide the building of several models as well as additional analysis following the results. The results of this study found that machine learning can be used to predict performance outcomes using display features as independent variables. Ultimately, however, the efficacy of such modelling was limited by the lack of inclusion of external variables that were presumed to impact the data. Such variables included: industry-specific user preferences, an increase in organizational preferences for multi-channel marketing strategies, the increasing complexity of the display ecosystem and finally independent author analytical capabilities

UddannelserMSc in Business Administration and E-business, (Kandidatuddannelse) Afsluttende afhandling
SprogEngelsk
Udgivelsesdato14 maj 2023
Antal sider84
VejledereLiana Razmerita