Targeted advertising is missing its target, research claims

Questionable results illustrate the pitfalls of personalized marketing.
article cover


· 3 min read

How well does targeted advertising work? We already know what audiences are giving up (personal info) in exchange for “relevant ads,” as the industry likes to say, but what are we getting in return?

A study by Adalytics, a browser extension that tracks and analyzes which ads a person sees, asks that question, and the answers are mixed.

  • For the study, researcher Krzysztof Franaszek tracked the browsing behavior of 25 volunteers over roughly two weeks. At the end, some were given surveys to gauge whether the ads served to them were relevant.
  • Insights were pulled from an ad’s clickthrough URL, which often show details around *why* someone’s being targeted, like their assumed gender or whether they’re travel-curious.
  • Although the sample size wasn’t large enough to establish definitive claims, the results illustrate the pitfalls of targeted marketing—and the potentially dubious data marketers rely on.

Case in point: 90.5% of ads for shoe company Merino were targeted to the wronggender, meaning men were seeing ads for women’s shoes and vice versa, the study found. And the National Rifle Association “repeatedly” served ads to two people it deemed “luxury vehicle enthusiasts,” even though neither said they had any interest in such cars (or in firearms).

Another user was served an ad for mattress brand Saatva that identified the user as “middle-of-the-funnel” for having searched the brand, even though the user had only searched the brand because they were given a hand-me-down mattress and wanted to check the dimensions.

The targeting data often comes from data brokers like LiveRamp or Epsilon, but because of the widespread use of third-party cookies, there’s a ton of junk out there that might be mislabeled or incorrect.

Get marketing news you'll actually want to read

The email newsletter guaranteed to bring you the latest stories shaping the marketing and advertising world, like only the Brew can.

As Franaszek told Marketing Brew, if you’re spending 400 million dollars, you don’t want $200 million being spent” on low-quality data.

Other takeaways

Amazon’s advertisements were seen by the most (64%). That’s not shocking, since Amazon is among the world’s largest advertisers, so it can afford to spend $$ to reach its desired audience, explained Franaszek.

Brandable Box, a custom packaging brand based in Atlanta, Georgia, reached 40% of the group on just three publishers: The New York Times, The Wall Street Journal, and The Washington Post. Because its ads only appeared on those sites, Brandable Box probably didn’t rely on programmatic advertising.

  • In other words: Buying ads directly from publishers—which is how advertising was bought 20 to 30 years ago—still works, said Franaszek.

Marketing 101 >>>

One of the study’s participants was served an ad for Ross University’s School of Medicine, despite not showing interest in medical school or planning to join the healthcare industry, even though the user was labeled “lower funnel.”

Another volunteer was shown ads intended for medical professionals—like on drug research—even though this person does not work in the healthcare industry. In this case, Franaszek said brands would have been better off investing in a relationship with, say, The New England Journal of Medicine instead of using programmatic advertising to try and '"find a doctor on a random website. Just practice basic marketing 101," he explained.

Zoom out: An entire ecosystem of advertising is predicated on collecting as much information about audiences as possible to serve personalized ads. Franaszek’s work asks a simple question: Does it work? “It does not appear to,” he said.

His advice? “Audit the quality of the data.”

Get marketing news you'll actually want to read

The email newsletter guaranteed to bring you the latest stories shaping the marketing and advertising world, like only the Brew can.