• by Laurie Sullivan , Staff Writer @lauriesullivan, July 29, 2024

New York Times Advertising is using generative AI (GAI) to increase the likelihood of successful results from ad targeting based on audience interests. 

The company announced BrandMatch, the tool, in early 2024 as an experimental GAI-powered targeting solution and began running beta campaigns with six notable brands across categories -- including tech, finance, luxury -- between April and May. The shift is paying off.

Joy Robins, global chief advertising officer at New York Times Advertising, told Data & Programmatic Insider in an email that the tool uses large language models (LLM) to support an advertiser’s brief and reveal more specific audience segments, and matches that brief to the most relevant articles to run ads against.

The team that designed the tool also built more than 200 proprietary audience segments using first-party data. It also relies on voluntary opt-in surveys and data modeling.

“We began offering early access to BrandMatch and opened the sale of media plans using BrandMatch on July 24, 2024,” Robins said, adding that the first campaigns are expected to be live in mid-to-late September.

Robins did not provide details of how the technology was implemented, but New York Times Advertising used a combination of third-party and first-party solutions to build the tool, which was built internally by a cross-functional team across New York Times Advertising and The Times’ ad product team. 

Granular targeting increase for luxury brand Ferragamo

Ferragamo was one of the beta testers that used BrandMatch to promote its summer collection.

The campaign served New York Times Advertising’s proprietary assets, FlexXLs. Ferragamo yielded a 0.81% click-through rate (CTR) when using BrandMatch targeting, outperforming benchmarks by more than 2X.

When measured through Adelaide’s Publisher Suite, the campaign yielded an AU (Attention Unit) score more than 40% higher than Adelaide’s global digital benchmark.

New York Times Advertising found that Ferragamo reached highly relevant audiences and content environments using the tool’s targeting and first-party measurement capabilities, with high CTRs.

Ultra-affluent readers — those with investable assets of more than $25 million — achieved a 1.22% CTR. Readers interested in Luxury Fashion achieved a CTR of 1.09%.

Fans of Paramount+'s "Evil" and/or supernatural genre

Paramount+ also participated as a beta launch partner. It used BrandMatch to promote the final season of its show, "Evil," with two target audiences: fans who already watched previous seasons of "Evil" and people who had not watched the show but are fans of the supernatural genre.

Using BrandMatch, Paramount+ submitted two slightly different briefs that described the subtle difference between the two audiences -- one that described why "Evil" fans like the show, and another that described the fan preferences of the supernatural genres.

BrandMatch tailored campaign delivery by finding articles that were likely to appeal to each audience, and modeling two unique audience segments based on people who read those articles.

Performance was measured through Adelaide’s Publisher Suite, designed for publishers.

The New York Times joined the launch in April. The campaign yielded an attention unit (AU) score 25% higher than Adelaide’s global digital benchmark, and had an overall CTR more than 20% higher than benchmarks for similar Entertainment advertisers using similar advertising formats during the same time frame.

Throughout the funnel, the campaign saw positive brand-lift results, with a lift of more than 2.2% in preference.

The performance between the two distinct audiences found the "Evil" fan audience outperformed CTR benchmarks by more than 35%, with peaks in CTRs driven by readers in the market for "Movie Theater Tickets" and entertainment subscribers.

The second brief was much broader, targeting "New York Times" readers with interest in Supernatural Thrillers -- a segment that outperformed CTR benchmarks by +14%. Peaks in CTRs were driven by articles that stirred emotion and audience interest in Awards season and those interested in the categories Horror and Thriller.

The tool, which advanced the New York Times Advertising’s ad-targeting capabilities, works in a different way than existing models.

“While traditional targeting models require explicitly choosing targeting criteria beforehand, this approach automatically interprets a brief and matches it with the most relevant context and audiences,” Robins said.