For Nimbler Ad Targeting, Meet ‘Alice,’ NYT’s Ad Library Service

In the past year, advertisers on The New York Times have bought ads with increasingly sophisticated targeting parameters, such as “adventurous” content, stories about to go viral or brand-safety keywords.

To make that in-depth content targeting possible, the Times built its own ad tech – Alice, shorthand for ad library service – a year ago. When a page loads, the Times calls the server-side ad library to analyze the content and return key-value pairs or targeting information about its sentiment, brand safety and virality.

The New York Times could have run such analyses on the page using JavaScript, but that would have added complexity that risked latency, broken scripts and broken pages. That approach would also require coordination across multiple teams covering Android, iOS, web and interactive content, slowing its ability to add targeting for advertisers.

“How do we do increasingly complex things across every platform, across 2,000 pieces of content a week and 150 million unique users without creating a mess?” said Allison Murphy, VP of ad innovation at the Times. “[Alice] was an elegant solution for that need.”

Because Alice doesn’t require changing code on the page, the ads team can easily turn different targeting parameters on or off and adjust them. For example, it could create a targeting rule for Olympics-related content for the duration of the event before removing it without changing any code on the page.

The ad library currently includes three main targeting parameters, all added in the past year since Alice launched.

Normally, it could easily take 18 months to integrate a single product because it would require coordination across 10 different teams. Due to the complexity, a rollout would occur a section at a time, just on desktop, for example, or only on mobile app.

“We went from a few standard offerings to launching three new distinct targeting capabilities,” Murphy said. “We could not have done this without the new ad library – it would have required so much resourcing.”

The ability to target content that inspires adventure, for example, comes from the data science initiative “Project Feels,” which predicts readers’ emotions after consuming content.

Advertisers that want to target viral content use an adaption of the tool Blossom, originally developed for the newsroom to understand social media response to stories.

Finally, the ad library houses Grapeshot’s contextual intelligence platform, which scans keywords for brand safety issues.

The ad library solves another problem for The New York Times: It reduced its reliance on third-party cookies, which are blocked by Apple’s Safari browser, to target ads based on user data.

“We know our content best,” said Pranay Prabhat, senior director of digital ad systems. “Having this solution in place means we can scale up and send the right targeting data [in order] to send the right message to users.”

In the future, Alice’s ad library will expand to run analyses of user information, not just content information. Improvements will roll out at a speedier clip due to its flexible setup.

“We can be in dialogue with advertisers about what they need and respond faster,” Murphy said.

This story has been updated to clarify that the Times still uses third-party cookies.


Enjoying this content?

Sign up to be an AdExchanger Member today and get unlimited access to articles like this, plus proprietary data and research, conference discounts, on-demand access to event content, and more!

Join Today!