NEW YORK (AdAge.com) -- Ads targeted at a particular context -- car ads on automotive sites, for example -- are a staple of online advertising. It's presumed that the more closely an ad matches a person's interest, the more likely that person is to to click and buy. And it couldn't hurt if the ad is big.
But a recent academic study indicates that may not always be the case. Indeed, a prominent targeted ad may have the opposite effect on consumers who perceive them as more creepy than helpful, or worse, an attempt to peel away their privacy.
"We were interested in the basic question of whether ads work better if they're targeted and more in your face, more visible," said Avi Goldfarb, a professor at the University of Toronto who wrote the paper with Catherine Tucker of MIT's Sloan School of Business. "We were expecting some kind of linear effect, that you would get a particular jump in effectiveness -- but the opposite is the case."
Either matching ads to a site's content or running obtrusive ads increases purchase intent, the study found, but doing both doesn't help much and sometimes hurts. When a more prominent ad unit such as an interstitial is also targeted, for example, the person seeing the campaign is only 0.3% more likely to intend to buy than if he or she sees a regular banner with no contextual relevance.
"You're basically not doing any better by doing both those things," Mr. Goldfarb said. "In fact, you're better off just doing one or the other, either highly visible placement or a contextual ad, but not both at the same time."
The noble banner ad
What's the most effective? The study found that simple banner advertising that mirrored the site worked best, where people were 0.9% more likely to buy than they would seeing a banner that had no contextual connection. For a much more noticeable placement without any targeting, users were about 0.5% more likely to buy than when they had seen a banner.
Mr. Goldfarb and Ms. Tucker suggest in their report that "advertisers could cut spending by over 5% without affecting ad performance," if they replaced specialized campaigns using both targeting and visibility with standard ads.
The report suggest that big units may make consumers think a little longer about the ads -- normally a good thing -- but might in the process give consumers a better chance to get spooked by the targeting. "Obtrusive ads may lead consumers to infer that the advertiser is trying to manipulate them," the study states.
Indeed, those concerned about privacy were turned off by such advertising. Among users who refused to provide personally identifying information such as household income, the study found that they were almost half a percent less likely to buy when faced with visible context ads.
"That has implications for more sensitive areas like finance and health sites," Mr. Goldfarb said. "For people who are more privacy sensitive, there was an overall negative effect. For them, these ads don't work at all."
"Our results show privacy matters in something of a subtle way in online advertising," Mr. Goldfarb continued. "Sometimes privacy violations are fine, sometimes they're not."