Targeting Decisions in Online Display Advertising
Answers from Prof. Dr. Nadia Abou Nabout
What’s your project about?
My research is about targeting in online advertising. The question of whom to target has been at the core of marketing, and targeted online advertising makes up a large portion of the global online advertising market.
What’s the research problem?
Advertisers know little about which of the hundreds to thousands of audience segments to select and test in their online advertising campaigns.
Which solution does your paper bring to the problem?
We develop a novel model to calculate the break-even performance of an audience segment, i.e., the performance necessary to make a targeted ad campaign at least as profitable as an untargeted one. Advertisers can use this break-even performance to decide whether to test specific audience segments in their campaigns (e.g., in RCTs).
How did you study this?
To show how advertisers can implement our model in practice, we use data from the Spotify ad platform and study the profitability of 71 audience segments.
What did you find?
We show that more than half of the audience segments offered on Spotify require an increase in performance that is larger than 100%. The literature on targeting suggests that such a high increase in performance is unlikely to be achieved for most ad campaigns.
What can practitioners learn from these results?
Using our model, advertisers can reduce the set of audience segments they wish to test for their campaigns to an amount that is possible to test. Instead of needing to test hundreds of audience segments (or thousands of combinations of them), advertisers can focus on the audience segments that require the smallest increase in performance and test those.
Get in touch with Nadia ABOU NABOUT to learn more about the project!