Here’s How Publishers Are Opening Their Data Science Toolkits to Advertisers

The New York Times recently rolled out machine learning programs

The New York Times recently opened two new data tools to advertisers. Getty Images
Headshot of Patrick Kulp

As publishers grapple with how to best make use of the troves of audience data at their disposal, a growing number are handing brands the keys to in-house data and artificial intelligence tools that could change the way ads and sponsored content are sold.

The New York Times, Group Nine Media and the Washington Post are among the media companies that have taken advantage of data science projects built for editorial purposes to give advertisers a clearer picture of who’s consuming their content and how to best speak to them. Publishers hope programs like these might help them gain back ground from tech giants like Facebook and Google that dominate the ads industry through targeting precision.

The Times debuted a unit earlier this year called nytDEMO that encompasses two new data-crunching tools. One, called “Project Feels,” is meant to gauge and analyze readers’ emotional reaction to articles and videos through a crowdsourced survey tool.

The other, Readerscope, tracks which topics resonate most with which readers on a more granular level than was previously possible and draws conclusions to better target advertising campaigns. Both tools were originally designed for the paper’s editorial and subscription sales departments.

Chris Wiggins, the Times’ chief data scientist, said the unit’s focus on data around the content it produces and how readers interact with it rather than precise audience demographics sets the project apart in an ad industry that largely trades on intimate knowledge of consumer identity.

“When you think about advertising and data–pretty much since the creation of the focus group advertising–data has meant data about people—for example, demographics—but that’s just a very small city within the continent of ways publishers use data,” Wiggins said. “There’s a lot we can do with machine learning and data and advertising that’s not necessarily the way people have understood the relationship between data and advertising for practically the last half-century.”

That means going beyond section and keyword targeting to more fully understand the various topics involved in the outlet’s coverage and the feelings they elicit. For instance, the company recently met with a social activism-focused brand to discuss a campaign designed to run specifically around content with characteristics that its tools have identified as “inspiring.” Another was offered space around media that exhibits “self-confidence.”

The Times declined to name any specific brands that have run ads informed by the tools.

The company has long used its internal R&D lab as a destination spot for advertisers looking to see what the news organization was building for its newsroom, with the hopes that products that start for the newsroom can end up in clients’ hands.

Indeed, the company opened up R&D Ventures, its commercial arm to its R&D Lab, in 2012 only to shutter it a year-and-a-half later. While Ventures didn’t last, the idea that a news organization can build products to help advertisers continues to push forward. A Times spokesperson said the nytDEMO team operates separately from the R&D lab.

Group Nine Media, the parent company of outlets like Thrillist, NowThis and The Dodo, for example, released a data science program at its NewFront presentation this month that it says will use AI to inform ad campaigns across its various brands.

Group Nine President Christa Carone said the tool, which it calls Group Nine Insights Analyst (GIA), has picked up on useful oddities in audience behavior, such as the comparative average watch time of dog and cat videos (dogs are watched 4.5 times longer) and a millennial predilection for spending long amounts of time on outer space content.

Until this month, such insights have only been used to inform editorial decisions, but the company will now open the door to advertisers. For instance, Group Nine recently worked with an auto brand on a campaign meant to persuade city dwellers to buy more cars that would run specifically around “road trip” videos that are three to four minutes long.

Elsewhere, the Washington Post announced a system called Own last fall that makes it easier for brands to syndicate AI-guided sponsored content and opened its Post Pulse trending topics tool to select brands. Bloomberg has also reportedly opened the doors of its data platform to advertisers as part of an initiative spearheaded by former Havas CEO Andrew Benett, now Bloomberg’s chief commercial officer.

Carone said she thinks advertisers will eventually come to expect advanced data and AI capabilities from publishers.

“I think that clients are going to start requiring this with their relationship with media–having a bit more insightful data that backs up the deal that they’re making and the returns on investment that they are hoping to get from their publishing partners,” Carone said.

Not everyone in the industry is quite so sure. Tony Bailey, Digitas’ senior vp of technology, said he’s seen more interest from brands in data science tools from ad networks than publishers themselves. Only a few major media companies have the sort of scale on their own to make proprietary tools worthwhile.

“There are a lot of people saying anything with an algorithm is AI,” Bailey said. “Most of the AI that’s out there right now is being used more in the buying space.”

@patrickkulp Patrick Kulp is an emerging tech reporter at Adweek.
Publish date: May 29, 2018 © 2020 Adweek, LLC. - All Rights Reserved and NOT FOR REPRINT