Why Behavior-Based Machine Learning Technology Is The Best Way To Attract (And Retain) Users
For almost a century, advertisers have helped keep the lights on across the media and news landscapes.
But as digital media outlets struggle to profit—and in many cases, survive—massive shifts are taking place across the entire industry. As evidenced by the long and growing list of shuttered media and advertising companies like Verizon’s Oath, advertising-focused profit strategies are no longer sustainable.
As someone who used to work for an ad tech company that was one of the many consolidated companies to become Sizmek, now bankrupt, I’ve seen this downfall firsthand.
In an effort to keep the lights on, companies across the digital publishing landscape are trying to avoid a complete dependency on display advertising and doing everything they can to gain more subscribers, which is no easy task. In this pursuit, digital news outlets are turning to machine learning as a potential subscription-driving savior. Real-time machine learning and predictive analytics can help by enabling organizations to deliver relevant content recommendations, which, in turn, drives subscriptions.
And major media players are already catching on.
Leading news sites like the Wall Street Journal and the New York Times have gotten serious about using machine learning to grow subscription revenues. While this approach has gained both traction and attention, most content providers have their work cut out for them.
Here’s why machine learning technology is the best way for any media company to gain subscribers and reduce churn on existing customers:
Most content recommendations are optimized for click-through rate—instead of focusing on what content can drive more subscriptions, which are much more valuable.
When judging the effectiveness of their content recommendations, major digital publishers often solely focus on attributable pageviews or click-through rate (CTR)—the ratio of clicks on your recommended content to the number of times each article is recommended. In this sense, clicks on the recommended articles serve as a reliable summary picture, especially as it pertains to recirculation.
But media companies today shouldn’t just be looking for clicks and pageviews—they need subscriptions to survive.
This is where machine learning can really help. Article recommendations drive traffic, but only with machine learning—which dives much deeper into behavioral data—can you determine what type of user experience drives subscriptions. Some articles simply drive more subscriptions than others, and that’s something machine learning will be able to predict much more accurately than an editor.
For example, a salacious article about Meghan Markle might be trending, but probably won’t lead to subscriber loyalty. On the other hand, an article from your paper covering local politics might actually motivate users to sign up for a subscription.
So while most content recommendations are only optimized to drive as many page views as possible, that's not the same thing as potentially tuning the recommendations to drive more subscribers.
Ultimately, curating a valuable user experience worth their time through data-driven article recommendations is the best way to drive subscriptions.
It’s simple: If users are reading more articles and visiting more pages per session, they’re more likely to subscribe—and then stay subscribed.
The real goal of machine-learning-based recommendations is to create value for users by delivering them personalized content they appreciate and will read.
Machine learning technologies like our product at LiftIgniter do especially well versus traditional recommendation systems when surfacing relevant evergreen and long-tail content, which helps outlets offer more value. Evergreen content is especially valuable because, with a great recommendation system, it can be shared over a long period of time with many users, thus maximizing return. Niche content—which is most valuable when distributed to the right users, and can really only be surfaced via machine-learning recommendation systems—creates loyal readers who come back to your website for that specific type of content.
Worth noting is that with technologies that use first-party data, publishers can track every visitor’s habits—not just subscribers’— through anonymous user IDs. This way, every visitor receives a curated experience. And, in turn, websites are able to put their best foot forward, enabling them to combat the infamous “Cold Start” obstacle—figuring out how to best understand and connect with new users—and maximize the value of each visitor.
The more your users read, the more value they’re receiving, and the more likely they are to subscribe or remain subscribed. In that pursuit, machine learning is your secret weapon.
Personalized emails keep your users coming back.
Once you’ve identified which content resonates, don’t stop at your website—email is a great way to keep them coming back.
The New York Times, for example, delivers personalized content to readers via newsletter emails called “Your Weekly Edition.” Clustering similar articles together in a neat package like this has proven to be a great subscription-driver. And this is something any media company can do with the help of a machine-learning technology like ours.
The night is darkest before the dawn.
It’s a dark time for the digital media industry. Every day, more newspapers and media sites shutter.
But with the power of machine learning—which can maximize reader value and drive subscriptions when implemented correctly—any media company’s prospects become much brighter.