The pivot to paid is bringing renewed focus on automated personalized content tools

Last year, The Times of London and Sunday Times developed its own self-learning algorithm to serve subscribers content most relevant to them based on their reading patterns. Early results showed the algorithm increases the likelihood that digital subscribers will open and click through on newsletters. Last July, Reuters redeveloped its app — although not for a subscription product — so users can customize the content they see. The success of the app is measured on total time spent in the app per month rather than pageviews. Also last July, The Telegraph introduced My Telegraph across its app and site, a feature where registered users can customize the topics they see, the journalists they follow and the save stories for later. The app is key in propelling it toward the milestone of signing up 10 million registered users in the next few years, after reaching its 2018 goal of 3 million in August.

The Telegraph wouldn’t reveal subscriber numbers but claimed its own app is a key driver in converting readers to registered users and subscribers.

“Our strategy has shifted toward getting people to come back to the website often, getting them to register for and become regular users of our services,” said Lucian Craciun, head of engineering and technology platforms at The Telegraph. “We recognize that people have an ever-increasing expectation of personalization from the content providers they interact with.”

Serving up content based on people’s preferences has weathered accusations in the past for being creepy or leading audiences into filter bubbles. According to the Reuters report, over 78 percent of people think it’s important to invest more in artificial intelligence but not as an alternative to employing more editors, hopefully laying claims to rest about robots stealing reporter jobs. This combination of human and machine in choosing content relevance is key. But as you’d expect, consumer concern has lowered as it’s become more commonplace.

“The risk comes when content that is peripherally linked gets presented in front of the consumer, the obvious question is how does that sit with my broader content set,” said Greg Harwood, director at strategy and marketing consultant Simon-Kucher & Partners. “Without some degree of human interaction, and in the absence of a clear consumer rationale, this can lead to confusion.”

Publishers like The Wall Street Journal, Norwegian publisher Aller Media and Swiss news publisher Neue Zürcher Zeitung have been experimenting with building flexible paywalls based on personal criteria, using propensity modeling and machine learning to spot patterns in conversion behavior.

“Every publisher has its own set of metrics around engagement and retention and is constantly running controlled tests,” said Harwood. “There is evidence of it driving engagement, but this is very individual and not necessarily in the public domain.”