Get Ready for Targeted Ads on Digital TV

October 5, 2017

Professor Carl Mela studies real-time television advertising

Professor Carl Mela studies real-time TV advertising

Online ads trying to sell us things we previously searched for are already the norm. But the advent of digital TV means our buying habits could soon influence the ads we see between our favorite television shows too.

“In theory, everyone could see a different ad, and that advertisement could be more relevant based on past purchase histories and viewing interests,” Professor Carl Mela said. “That’s a radical shift.”

Mela, a marketing professor at Duke University’s Fuqua School of Business, reviewed the viewing habits and purchase histories of 834 households across an entire year to study how advertising could be more accurately and profitably targeted.

“As television becomes digitized, capabilities are beginning to appear that never before existed,” Mela said. “We leave digital footprints when we watch television. Every keystroke of your remote could be recorded. So for the first time firms can try to understand the viewing preferences of individual viewers and communicate with them one-on-one.”

"For the first time, we have household-level viewing information to understand about how people view TV."

Mela drew viewing information from set-top box usage data, purchase data from the market research firm Information Resources Incorporated, advertising rates from TNS Research, and national viewing data from AC Nielsen. All data were from between June 2005 and July 2006.

“For the first time, we have household-level viewing information to understand about how people view TV,” Mela said.

His findings, “TV Viewing and Advertising Targeting,” are forthcoming in the Journal of Marketing Research. Mela worked with Yiting Deng of University College London, who received a Ph.D. from Fuqua.

The researchers found most people tend to watch television every night, and for most of the evening. They looked for patterns of ad avoidance – when and how viewers were more likely to skip ads, through changing the channel or skipping through them on recorded shows.

What they found – hidden until now by the lack of data on individual viewers – is that rather than there being certain times or shows during which viewers were more likely to skip ads, most of the variation in ad viewing was between people. Some are simply more likely to avoid advertising than others.

“Most of the variation we find in ad skipping is at the individual level,” Mela said. “Advertisers should not buy exposure to people who avoid watching them. That’s intriguing because it suggests there’s a considerable potential from targeting ads across individuals instead of shows, which is the current practice.”

“If an advertiser knows you’ve been exposed to ads in the past and can link that information to your purchase behavior, it becomes possible to understand whether you’re responsive to advertising."

More importantly, combining the various streams of data allowed the researchers to see which viewers respond to ads.

“If an advertiser knows you’ve been exposed to ads in the past and can link that information to your purchase behavior, it becomes possible to understand whether you’re responsive to advertising,” Mela said. “The real value of having single-source data is the ability to target people who are most interested in ads.”

This could enable firms to target heavy consumers of peanut butter who frequently switch brands, for example. Or auto companies to show car ads to viewers who own older vehicles.

“Digital television advertising is going to be a substantial market,” Mela said.

The findings could also be applied to users of Netflix and similar services, Mela said.

“Those platforms could use it to optimize their show scheduling, assortment, or recommendations,” he said.

To protect privacy, the information harvesting takes place through a process called double-blind matching, Mela said.

“The cable company only knows what you watch and the ad company only knows what you bought,” Mela said. “The complete picture assembles around a random ID number so no firm has all the data. No advertiser knows what you’ve personally been watching.”

Mela said better targeting of ads is good for consumers as well as firms.

“Making this whole process more efficient is better for consumers and better for companies,” he said. “It’s a waste of resources and a drag on the economy when men are seeing ads for women’s razors and women are seeing ads for men’s razors – gifting aside, of course.”

Contact Info

For more information contact our media relations team at media-relations@fuqua.duke.edu.