To ape Charles Dickens, it is the best of times, it is the worst of times — for media today. Even as President Donald Trump assails news coverage of his administration, his election has reminded America’s citizenry of the value of a vigorous and independent press. And the standard-bearer for speaking truth to power is, in many ways, the New York Times.
The fabled newspaper is receiving a perhaps unprecedented level of both scrutiny and praise — which translates into soaring levels of traffic and reader engagement and presents a steep challenge for CTO Nick Rockwell. He’s tasked himself and his team with using machine learning to provide what he calls “an indispensable reading experience, leading to a high propensity to subscribe.”
And it’s working.
Rockwell described what’s afoot with AI and the New York Times in a conversation with VentureBeat editor in chief Blaise Zerega today at MB 2017.
“We track everything we can think of — everything a user reads,” said Rockwell. “We look at email, marketing they have responded to, where they are coming from, events we are interested in that is valuable to us. We generate a fair amount of data.” Rockwell added that he and his team pay close attention to preserving serendipity and discovery, not in an anachronistic way, but in one that meets readers’ needs and expectations.
“There are two core problem we are trying to solve — creating a relevant rapport with readers,” Rockwell said. “It is a long process of introducing people to what we do. Alongside that is understanding the journey [to become a subscriber] that will promote the conversation to fund our newsroom. Both of those are tractable to data analysis and machine learning techniques.”
Even as President Trump publishes scathing tweets of the “failing NYT,” the old gray lady continues to report strong growth in its digital subscriptions and profits. The company has reported a steady surge in new online readers, having added 276,000 subscribers in the fourth quarter of 2016 and another 308,000 in the first quarter of this year. This has helped drive 11 percent growth in circulation revenue and a 19 percent jump in digital advertising revenue for the most recent quarter. The company says it currently has 2.2 million digital only subscribers.
Rockwell uses analytics and AI to develop new personalization strategies, measure their performance, and then improve them. This means understanding the way 150 million unique visitors interact with the thousands of articles, videos, and blog posts published each month. People generally enter the nytimes.com site either by navigating to the homepage or by going directly to specific articles, using referrals and search.
“We are producing 200 to 300 pieces of journalism per day,” Rockwell said. “We know from the data that experiencing the breadth of reporting is one of the main things that leads to a propensity to subscribe.”
Much like the newspaper’s front page, a great deal of editorial thought goes into what needs to be on the homepage, Rockwell explained. But it’s after the initial click or article read that personalization strategies become paramount.
Traditionally, New York Times readers might have sampled the newspaper’s printed edition in their 20s before becoming subscribers in their 30s. Rockwell said that with the overall growth of online news and the so-called Trump Bump, this conversion timeframe has fallen to between 18 and 24 months.
“We’re susceptible for good or ill to the news cycle,” Rockwell told VentureBeat in an earlier conversation, explaining that AI is critical to helping his company weather the ups and downs of Trump, and more. “Machine learning is helping us discover what baseline behaviors are most important. Subscriptions are now the core of our business. It’s our stated strategy.”
Additional reporting by John Brandon.