Project Feels: How USA Today, ESPN and The New York Times are targeting ads to mood

“We’ve seen a gradual increase in RFPs, advertisers aren’t asking for audience by demographic but psychographic,” Andresen said. “This is one step to find those psychographics instead of numbers you can get from outside parties.”

The New York Times rolled out a tool earlier this year called Project Feels that lets advertisers target ads to content based on emotional responses the content is predicted to have. ESPN has been pitching a tool to target sports fans on its digital properties based on their changing emotional state during a game. Now it’s trying to apply all that know-how to the rest of Disney’s properties. The tool is called LiveConnect and takes the sports preferences that its logged-in users provide and overlays that with data about how people feel based on how their team is doing and line that up with advertisers’ goals. That could mean showing ads for travel or other celebratory experiences to people whose team is on a winning streak — or not advertising to them at all if their team is losing.

“We were trying to find any scalable way to deliver the right message at the right time,” said Vikram Somaya, svp, global data officer and ad platforms at ESPN. “Sports, there’s a lot of potential messages around it. The fact is, we’re in the emotion-generating business. We have a lot of first-party data we can use with a lot of transactions with the consumer; why wouldn’t we do it?”

“It’s something a lot of media planners have been trying to do for years,” Chris Wexler, svp and executive director of media and analytics at Cramer‑Krasselt. “It just depends on how rich the data set is. It’s an exciting frontier because we’re looking for people who are open to our message, and emotional state is a key part of it. Demographics based targeting is better than no targeting. Behavior-based is better than demographics. If you put it in order of importance, mood is above behavioral.”

Hitting people with a message when they’re likely to be receptive is as old as advertising, but using artificial intelligence to target people based on their mood is another level of manipulation. Publishers and agencies say they draw the line at using personally identifiable data or data without saying what it’s being collected for. “This is really about focusing on finding better-qualified audiences, but this is not at a point where we can manipulate mood through content,” Andresen said. ESPN’s Somaya said it won’t use information it thinks people don’t feel should be publicly revealed, keeping in mind its family-focused Disney audience, which means everything from health to political preferences to sexual orientation. That data isn’t as commercially valuable as information they willingly provide (like sports interests) anyway, he added.

The biggest challenge is proving mood-based targeting works and at scale. It’s still just an “occasional” part of the ad sell at publishers. It takes a client that’s willing and able to work closely with the publisher on figuring out what the ROI should be, measuring it, and spending the premium that’s often charged.

Agencies also want more information about the methodology. It’s an easier sell with music because it’s easier to believe as companies like Spotify might say that people who listen to break-up music are sad.

Josh Baines, who works on publisher strategy at Oracle, whose Grapeshot product powers mood-based targeting, said the topic is getting more visibility because it provides a deeper level of targeting and chance to qualify people’s mindsets. Showing it drives real business outcomes and not just fluffy metrics will be key to growing its adoption, though, he said.

Other categories outside of music are a harder sell. Some advertisers shy away from being around news because they think it will create a negative association with their brands with readers. For news publishers, one hope of A.I. is to change advertisers’ associations to news. Andresen said USA Today Network found that people are equally drawn to “positive” versus “negative” stories when they come to the site, a finding it uses in pitches to advertisers. “It’s helpful for us to explain that to advertisers: You don’t have to walk away from the whole news category.”

Wexler isn’t so sure. “In music, they can claim to know when you’re sad. We’d be particularly skeptical about news content — are we going to guess that because someone’s in Alabama they’re angry about something versus someone in New York?” he said.