Prisa Media, publisher of news titles including El Pais and Cinco Dias, is using machine learning to develop emotion-based segmentation in order to refine the context of hard news stories that advertisers might otherwise be reluctant to advertise against.

“News can’t be happy all the time; that’s the reality,” Pedro Ventura, director of technology on data and monetisation at Prisa Media, told Digiday.

“We’re general news, we cover a broad spectrum, and, of course, much of that is terrorist-related news or the Catalonia conflicts. Some advertisers just don’t want to be near that.”

But not all hard news falls into this context and the publisher is in the process of building an algorithm that will enable it to better distinguish how readers respond to the stories it carries and decide whether to offer ad impressions around the article.

Having identified among the 15,000 articles published every month on El País those that advertisers might shy away from, the publisher has worked with a market research firm and 2,000 participants to log reactions, with this data fed into the algorithm.

“Having that human element is important in order to train the algorithm,” Ventura noted – and this training will continue for another year.

Off the back of this, Prisa Media has created 32 audience emotions and a number of segments, e.g “happiness”, that it plans to offer to advertisers at a higher price than its regular targeting options, some major brands are reported to have shown interest, said Digiday.

Past research by Yahoo has shown that advertisers able to reach consumers when they’re in the right mood could significantly increase the impact of digital advertising: when consumers were upbeat they were 24% more receptive to content in general, but 40% more likely to be receptive to digital advertising specifically.

If Prisa Media is successful with its new tool, it aims to widen the scope of the algorithm to take in other languages, such as English and French, and to sell it to other publishers. Earlier this month, the New York Times revealed that it too was working on a form of “perspective targeting”, which allowed the publisher to place ads based on the emotions certain articles were likely to evoke. 

Sourced from Digiday, Poynter; additional content by WARC staff