Why you’re still getting irrelevant ads, according to research

Tired of getting followed by ads for things you don’t want? New research from the University of Florida explains why this happens and how advertisers can help address irrelevant ads.

GAINESVILLE, Fla. – Imagine you’re in search of a new car. You’ve seen a Subaru model that interests you, so you decide to look it up online for more details. After some brief research, you quickly decide the Subaru isn’t for you. Instead, after additional searching, you decide that it’s a Toyota model that fits your needs. You continue to research your chosen Toyota.

Despite closing the door (and browser tab) on the Subaru, and spending significantly more time browsing the Toyota information, you seem to only be getting reminders of your Subaru browsing history in digital ads.     

This scenario describes an all-too-common problem that digital advertising promised to solve. Leaving behind the spray-and-pray tactics of the past, digital advertising claimed to be a way to target customers more efficiently.

Online users know better, though. In fact, in a survey of 287,000 mobile users, a 2019 report found that 90% of them considered targeted mobile ads to be annoying, up from 78% in 2018. 

Despite upgrades in targeting technology and a wealth of access to consumer data, why haven’t you been able to shake irrelevant ads? Research from the University of Florida Warrington College of Business finds that ad agency incentive structures are part of the problem.

“Recent developments in ad technology have enabled ad publishers to sell individual consumer impressions (eyeballs) as they arrive in the website,” said Woochoel Shin, Brian R. Gamache Professor and co-author of the research. “This process potentially leads to an efficient delivery of relevant ads but ad agencies, in their effort to deliver the promised number of ad impressions to their clients, may distort the efficient match.”

 Shin and his co-author Jiwoong Shin from the Yale School of Management used a model to replicate the ad purchasing process for an ad agency serving multiple clients. With each client, the ad agency is contractually obligated to allocate an impression, or an ad view, but are limited by the number of ads available in the market and each client’s budget. In the common scenario in which one advertiser is more mainstream than the other, an agency faces a tradeoff.

“It can allocate the first impression to the mainstream advertiser who, generally, has a higher match probability,” Shin writes. “Or it can save this advertiser for a future impression, which may have an even higher match probability. This tradeoff arises because the ad agency needs to serve both advertisers who have a budget constraint.”

The researchers find that when the asymmetry between two advertisers is either extremely large or small, the ad agency is more likely to allocate the ad impression to the more niche advertiser, whose match probability with an impression is generally low. It’s in this scenario that you see more irrelevant ads across your digital platforms.

“The agency essentially ‘dumps’ the current impression to a less relevant (or even completely irrelevant) advertiser,” Shin explains. “The agency’s strategic consideration creates this inefficiency, manifested most obviously in completely irrelevant advertising for consumers even in the presence of perfect targeting accuracy.”

While the advertiser doesn’t have complete control over which audiences see their ads when working with an ad agency, it can play a role in helping address irrelevant targeted ads. Shin and his co-author recommend that it begins with advertisers recognizing the distinction between agency-induced irrelevant advertising from error-based irrelevant advertising.

“For example, advertisers may choose to contract with the agency that also serves advertisers with a similar level of appeal to the consumer population or even avoid the agency serving other advertisers with potentially overlapping consumer segments,” Shin shared.

This research is forthcoming in Management Science.