Machine learning: Powering advertising and audience

By Scott M. Scher, Columbia University

“Independent News Media has fueled Ireland’s news conversation since 1908. A lot has changed from Linotype to digital, but the trust between IMN and its 1 million daily online users remains as strong as ever,” said Brendan Hughes, chief digital officer of Independent News & Media in Ireland.

The audience that IMN reaches is varied, working mothers, baby boomers, and millennia’s, yet it remains the most engaged compared to its competitors. The audience was dazzled with stats like 33% more reach to 50+ users, 27% more reach to globe trotters, 25% more reach to females, and 21% more reach to main shopper with kids. This does not mean that IMN does not have a challenge they need to meet. That challenge was expressed as “identifying the audience segments that perform for our brand advertisers”.

Using AI and machine learning IMN finds these segments in three ways. First, contextual- align brand stories with the content that matters most to its audience; second, behavioral- using first party data to target the most desirable and engaged audience segments; and third, social- amplify brand messages through demographic and interest targeting on social channels.

The goal of IMN is to “develop a scalable approach to analyzing first-party audience data to identify segments with high intent around specific categories.” They intend to demonstrate that using their intent full audience segments actually improves campaign performance.

Using the AI platform HeyStaks and working with a travel client they created a campaign that targeted their travel reading audience and did an A, B split. They took 50 percent of its users and put them through the HeyStaks platform to do a broader profiling on the users looking at all of their content activity across different times of day and across devices to identify patterns of behavior that transferred over user groups that would help find correlations for what they call “act-alike” audiences. At first the behavioral segment and the AI segment had about the same level of engagement, but the AI group increases significantly over the course of the campaign. They were 3x more likely to engage, 33x more audience members, and 5x more members.

With this data they were able to provide new sales benefits, drive performance, educate the market on the internet, automate the process, and combine first party data into a single digital mapping platform.