Author Information : Pengyuan Wang (University of Georgia) Guiyang Xiong (corresponding author; Syracuse University) Jian Yang (Oath Inc)
Year of Publication : Journal of Marketing (2018)
Summary of Findings : Compared to traditional undisguised ads, native ad (a new type of online ad) effectiveness changes in a unique pattern as ad serial position lowers.
Research Questions : How quickly does a native ad’s performance change as its serial position lowers (e.g., from rank 1 to a lower rank)? How does the rate of change vary for different performance metrics and viewer groups?
What we know : This study adds to the literature on ad serial position, which has focused on undisguised and interruptive ads. We show that the effect of serial position exhibits a unique and interesting pattern for native ad, because of its disguised and non-interruptive nature compared to traditional ads. Specifically, as native ad serial position lowers, there is only a moderate reduction in publishers’ metrics, but a drastic reduction in advertisers’ metrics. Such vastly asymmetric effect of serial position has not been documented by prior research on other types of ads. Moreover, we show the contingency effect of native ad serial position, e.g., how it varies across audience groups and webpages. There is little empirical literature on how consumer characteristics and webpage contents moderate the effect of serial position. Our findings are of managerial importance to both advertisers and publishers. We show that the prevalent ad ranking system places native ad advertisers at an unfair disadvantage (advertisers overpay for lower rank positions). Moreover, we provide new insights about how advertisers and publishers can improve native ad performance under different conditions (e.g., for different consumer groups and different webpages).
Novel Findings : As serial position lowers, the performance of a native ad drops only moderately for publishers -- in terms of click-through-rate (CTR) and revenue per impression, but acutely for advertisers -- in terms of average conversion rate (CVR) and conversion per ad dollar spent. Moreover, the moderating effects of audience gender and age are asymmetric for publishers metrics versus advertsers' metrics.
Novel Methodology : Prior research on ad serial position largely relies on lab experiments, and the limited empirical studies typically analyze data from one particular advertiser and may lack generalizability. In contrast, we examine large-scale field data on 120 native ads covering over 180 million ad impressions. For each ad, we use separate “natural experiment” studies to compare its own performance as its serial position varies. Subsequently, we conducted a meta-analysis to generalize the results across all separate studies.
Full Citations : Pengyuan Wang, Guiyang Xiong, and Jian Yang (2018), "Serial-Position Effects on Native-Advertising Effectiveness: Differential Results across Publisher and Advertiser Metrics", Journal of Marketing, forthcoming.
Abstract : The advertising industry has recently witnessed proliferation in native ads, which are inserted into a web-stream (e.g., a list of news articles or social-media posts) and look like the surrounding non-sponsored contents. This study is among the first to examine native ads and unveil how their effectiveness changes across serial positions by analyzing a large-scale dataset with 120 ads. For each ad, we use separate “natural experiment” studies to compare its own performance as its serial position varies. Subsequently, we conduct meta-analysis to generalize the results across all separate studies. The results reveal vastly asymmetric effects of native ad serial position on publishers’ metrics (click-based) versus advertisers’ metrics (conversion-based). As serial position lowers (i.e., from rank 1 to a lower rank), there are only modest changes in publishers’ metrics, but drastic reductions in advertisers’. This pattern is unique to native ads and has not been indicated by prior research on ad serial position. Moreover, we show the moderating effects of audience gender and age. The findings provide new and timely implications for researchers and marketers.