The future of A/B testing in social network advertising

Abstract

This research addresses the future of A/B testing in social network advertising. A/B test is a well- studied comparison problem with two different samples with the goal of testing the treatment effect of old and new variations. In recent years, through the rise of the internet, A/B testing in social networks has gained sharpened focus and is commonly used in social network advertising. Due to the market-driven strategy the companies should today aim for, the development of A/B testing in social network advertising can help in gathering useful insights of consumer preferences and attitudes. A/B testing has been perceived as cheap, simple and reliable way of optimizing advertisement and mining data from site users. However, as currently performed A/B testing has criticized as manual and time-consuming activity that requires complex set of statistical and engineering skills. This study focuses on overcoming these problems through automation and machine learning algorithms. Besides, the importance of shifting organizational focus on optimal usage of data-driven decision making through A/B testing, and user attitudes towards social network advertising and their ad-clicking behaviour are addressed.