Decision sciences professor Ali Hojjat knows how to optimize online ads

Wednesday, March 28, 2018
Associate professor of decision sciences Ali Hojjat

Associate professor of decision sciences Ali Hojjat studies online advertising and optimization

Ever noticed those ads that pop up or fill in the banners and borders of the web pages you visit? Advertisers, it goes without saying, are hoping you do.

Internet ads may be annoying but they’re an inescapable part of everyday life online. They also comprise a $72 billion industry that reaches many millions of people each day. At any given moment, thousands of brands are running online campaigns seeking to engage us. The big online publishers drawing these eyeballs together—Yahoo, Facebook, Google, etc.—must satisfy advertisers that they are delivering a desirable number of impressions to a carefully defined demographic.

If advertisers could find a way to improve the efficiency of ad delivery by even 1 percent, they could generate millions in additional revenue. Enter Ali Hojjat’s algorithm.

“Currently, online advertising is impression based. Web sites promise advertisers that a certain number of ads, or ‘impressions,’ will be served to a defined demographic. It doesn’t matter how many times each individual user will see the ad as long as the total impression count is reached,” said Hojjat, assistant professor of decision sciences in the Peter T. Paul College of Business and Economics.

Hojjat’s research has uncovered a way for online publishers to efficiently deliver campaigns with specified reach and frequency.

“Now advertisers can specify how many unique individuals should see the ad, and how many times each individual user should be exposed,” he said.

Hojjat’s algorithm is tailor made for cutting-edge industry trends, such as storyboarding—advertising messages over a sequence of images or videos that tell a brand’s story. It allows advertisers to ban competitors from appearing to the same user they are targeting during a campaign and/or specify how ad impressions should be paced over time for each user.

The algorithm utilizes the growing dominance of user interaction with mobile devices that has enabled online publishers to accurately track and model the browsing behavior of viewers over long periods of time.

Ali has tested his algorithm at Yahoo Research, but said that social media platforms such as Facebook, which require a user login to interact with the website, would benefit significantly more from such an algorithm as “they have a near-perfect visibility and tracking capability on individual user visits.”

Perry Smith | Freelance Photographer