CNN’s data-powered audience development strategy

By Scott M. Scher, Columbia University

CNN thinks about data in a mission driven way unlike other “algorithms gone wild” organizations. Christopher Herbert, SVP of operations and strategy for CNN Digital, equates this to parenting or “wielding the power of data in a thoughtful way”. CNN mission statement: to be the world’s most essential and engaging news source. CNN uses three metrics to measure if they are achieving this mission; daily unique visitors, video starts, and frequent visitors.

Herbert explained to the audience that CNN sees data as a means to enhance, grow, and understand its environment, as well as a means to build new products for its customers. What they do not do is, feed their goals into an algorithm and allow a program to optimize for these goals and select what goes into feeds, or tell writes what stories to write. They do not compromise their journalistic standards trying to achieve their mission.

An example of this is how AI allows CNN to create topic modeling to filter out the “signal from the noise” amongst reviews the CNN app receives. According to analysis conducted through AI, newsletters are CNNs highest traffic content and with the use of adaptive diagraming have been able to increase its users video consumption, create fresh content, and develop content targeting.

Through Machine Learning and AI, CNN is able to understand and tag context within its articles, and create new ingredients for success, such as creating the strongest team, putting that team in the right place in the organization, and developing the right tools for its job.

Within CNN, AI is a tool that allows the organization to foster a culture of testing and experimenting in the hopes of creating the best outcomes and goals. To the employees of CNN, data is greater than opinion and goals are clear, measurable, and linked to strategy. Herbert believes that AI holds the key to enhancing and expanding human judgment.

AI analytics has provided CNN with tangible evidence of how their customers are interacting with their homepage. The most clicks they were receiving were a result of their newsletter, indicating a high level of engagement back to the site. This analysis is used to evaluate where CNN looks for its users, making sure they look in the right places. They use AI to evaluate how they onboard their customers and this informs their production strategies. Adaptive program looked at how to build up their video consuming users. At first they looked at isolating their high video consumption users and showing them more videos on the homepage as a means to increase video consumption.

As they tried to build out that heavy video user base “It turned out that was less of a thing then one would think and what was more of a thing was frequent users”. Users that came back more frequently were exposed to more chances to watch videos and that was a more important thing to look at. This created a new question, “how do we keep our content fresh for frequent users?”

Herbert showed a graph of homepage activity which showed on the first day for high frequency visitors how many times they clicked on the homepage. Sixty percent of users on their first time for the day visiting the homepage clicked once, 20 percent clicked twice, and then clicks fall off from there. On the second visit the number who clicked twice fell slightly, and the amount of people who didn’t click but just skimmed the page increased from about 5 percent to 10 percent.

By the time you get to the fifth visit, the amount of users not clicking at all has jumped up to 90 percent. In an effort to give users a fresher experience and drive traffic, CNN keeps its homepage fresh by moving zones around, changing some zones, and injecting new zones. This strategy has driven about a 30 percent increase in engagement on the homepage.