Skip to main content
5 blocks, with one block in the middle showing a green target on it

As the Accuracy of Targeted Advertising Increases, So Do the Pitfalls

New research from Yale SOM professor Jiwoong Shin highlights how the predictive power of targeted advertising can be undercut by ignoring the misaligned incentives of advertisers and consumers.

Marketers understand by now that when it comes to shelling out for ads, it pays to target the consumers most likely to buy their products. Study after study has established that the effectiveness of advertising increases when it is strategically directed—something that advertisers can do with increasing precision thanks to online data collection.

As advertisers’ targeted-ad algorithms improve, one might expect that their laser-focus would become even more fine-tuned and profitable.  But while targeted-ad algorithms have improved, the regulatory environment and the increasingly present reality of the so-called “cookieless future” has created doubts for how targeting can be utilized to increase profitability. Yale SOM professor Jiwoong Shin, echoes these doubts. “This [fine-tuned targeting and increasing profitability] is the core promise for the bright future of the digital advertising industry,” he writes in a new paper co-authored by Z. Eddie Ning, assistant professor of marketing at the University of British Columbia, and Jungju Yu, assistant professor of marketing at the Korea Advanced Institute of Science and Technology. “However, the reality is not necessarily playing out so well.”

In their paper, the researchers explore the ways in which targeted advertising can backfire. Since the arrival of regulations (like the EU’s GDPR and California’s CCPA) that allow consumers to limit the tracking of their online lives, advertisers can more easily undermine their own efforts to maximize profits—by unwittingly pushing the consumers they’re targeting to opt-out of data collection.

To study how this might happen, the researchers built a model to simulate online interactions among a firm advertising a product and a group of potential consumers.

One of their operating premises was that many people actually interpret targeted ads as personalized recommendations, a tendency that previous research has established. Advertisers are, of course, aware of this tendency. Given that their own motivation is maximizing sales (and not necessarily offering thoughtful recommendations for their own sake), the researchers conjectured that advertisers would be incentivized to take advantage of consumers’ trust in the personal match of products featured in their targeted ads.

This is exactly what their model revealed. “Given the misaligned incentives, the firm’s better prediction capability does not necessarily translate into more accurate targeting,” the researchers write.

The model showed that advertisers were likely to take advantage of consumers’ trust in the accuracy of targeted-ad prediction by “cheating” and broadening the ad’s viewership, sending it to people who were assumed to be weak matches with the advertised product. The advertiser was simply betting that even some of these weaker matches may buy the product, an outcome made more likely by their seeing the ad as a recommendation.

At least, this was what happened when the researchers modeled a scenario in which product prices were kept fixed. Next, they modeled a slightly different scenario: one in which product prices could fluctuate depending on the firm’s confidence in the accuracy of its predictions. In this case, the advertiser was less likely to send ads to weak-match customer and instead simply increased prices for best-match customers.

Unlike the scenario with fixed product prices, this one was likely to backfire for advertisers. As consumers confronted higher prices, they were driven to protect their privacy and disable ad targeting. As a result, advertisers were left empty-handed, without the option to target those potential consumers at all.

“These results illustrate the limit of prediction technology in reducing market friction,” the co-authors write. Advertisers would do well to be aware of the misaligned incentives dictating these online interactions around ad-targeting, clicking, buying, and opting-out. “As privacy laws go into effect, consumers’ potential decision to opt-out may discipline the advertiser’s incentives, ultimately influencing the advertiser’s targeting strategy and consumer welfare.” The resounding result of the research: advertisers should err on the side of caution with the use of re-targeted ads if they want to maximize profits.