@article{a12d94f003954fa3ac89015f74be849f,
title = "Bypassing Performance Optimizers of Real Time Bidding Systems in Display Ad Valuation",
abstract = "Demand-side platforms (DSPs) that purchase digital ad space using real-time bidding (RTB) systems employ “black-box” performance optimizers to adjust bids at run time. Advertisers using field experiments to estimate the marginal value of display ads need to contend with the selective targeting of users by optimizers that adjust bids to target users with a greater propensity to respond favorably (i.e., click or conversion). In this paper, we propose an alternative approach for advertisers who choose to bypass their DSP{\textquoteright}s performance optimizers for the purpose of assessing the value of their ads. We show that external frequency caps that set upper limits on the number of ad impressions outside the purview of bidding algorithms can serve as a suitable instrumental variable. Eliminating performance optimizers allows the advertiser to value ads without relying on the support services of the DSP with the added benefit of a broader customer reach and a markedly lower cost. As the focal advertiser disables performance optimizers, any overbidding or underbidding vis-{\`a}-vis competition that employs them results in a negative correlation between the numbers of ad impressions won and their underlying quality in real time. Using two large-scale randomized field experiments in different geographies (United States and Asia) and different devices (PC and mobile), we validate the proposed approach and report a positive effect of ad impression count after adjusting for net negative bias.",
keywords = "advertising effectiveness, digital display advertisement, field experiment, frequency caps, programmatic advertising, real-time bidding",
author = "Christopher, {Ranjit M.} and Sungho Park and Han, {Sang Pil} and Kim, {Min Kyu}",
note = "Funding Information: This study is financially supported by the Institute of Management Research at Seoul National University. The authors express their gratitude to the senior editor, Param Vir Singh, and the associate editor, Vibhanshu Abhishek, for their crucial insights and support during the review process. The authors also thank Bradley Fay for assistance with data collection and liaison with one of their industry partners pertaining to the preliminary field experiment. Ranjit Christopher acknowledges valuable feedback on the early versions of this manuscript from the attendees of the 2017 INFORMS Marketing Science Conference and the attendees of the 2018 Research Colloquium at the University of Missouri–Kansas City. Sungho Park acknowledges generous support from the Institute of Management Research at Seoul National University. Sang Pil Han acknowledges Mathpresso, Inc., the developer of QANDA, for their generous support of artificial intelligence research for the digital economy. While conducting this work, Sang Pil Han served as an advisor to Mathpresso, Inc. Funding Information: History: Param Vir Singh, Senior Editor; Vibhanshu Abhishek, Associate Editor. Funding: This study is financially supported by the Institute of Management Research at Seoul National University. Supplemental Material: The online appendix is available at https://doi.org/10.1287/isre.2021.1050. Funding Information: The authors express their gratitude to the senior editor, Param Vir Singh, and the associate editor, Vibhanshu Abhishek, for their crucial insights and support during the review process. The authors also thank Bradley Fay for assistance with data collection and liaison with one of their industry partners pertaining to the preliminary field experiment. Ranjit Christopher acknowledges valuable feedback on the early versions of this manuscript from the attendees of the 2017 INFORMS Marketing Science Conference and the attendees of the 2018 Research Colloquium at the University of Missouri–Kansas City. Sungho Park acknowledges generous support from the Institute of Management Research at Seoul National University. Sang Pil Han acknowledges Mathpresso, Inc., the developer of QANDA, for their generous support of artificial intelligence research for the digital economy. While conducting this work, Sang Pil Han served as an advisor to Mathpresso, Inc. Publisher Copyright: {\textcopyright} 2022 INFORMS",
year = "2022",
month = jun,
doi = "10.1287/isre.2021.1050",
language = "English (US)",
volume = "33",
pages = "399--412",
journal = "Information Systems Research",
issn = "1047-7047",
publisher = "INFORMS Inst.for Operations Res.and the Management Sciences",
number = "2",
}