Attracting subscribers is one thing, preventing them from leaving is another. Enter ‘churn prediction models’ which are becoming increasingly effective at reducing churn rates by identifying key behaviour patterns in advance. The Data Business explains more…
Businesses of all sizes have long used forecast models to gauge when their clients might terminate their contracts. With the rise of big data, media organisations are increasingly adopting this practice to great effect. The challenge is predicting when it’s going to happen and reducing it.
Data scientists at a number of news publishers try to identify signals and patterns that pinpoint customers who are least likely to renew their subscriptions. Known as ‘churn prediction’, these models help companies develop strategies to retain subscribers and forecast revenue for the year ahead.
Two news organisations that have successfully adopted these models are the Norwegian newspaper Aftenposten and The Wall Street Journal.
Using data to prevent churn
Norwegian newspaper, Aftenposten, developed a churn prediction model in 2015. This enables the newspaper to identify readers who are likely to churn, so they can then target them with retention measures like special promotions and discounts.
The churn prediction model combines several data sources like subscription type, customer demographics, and how and when the subscriber was in contact with Aftenposten. Their data team is then able to identify indicators that are strongly linked to churn, for example, low digital engagement, days since purchase, and manual payments with short renewal periods.
The data is then used by the customer support department to contact subscribers who are at risk of cancelling their subscriptions and try to keep them onboard. Before implementing this data strategy, nearly 60 per cent of Aftenposten’s income came from advertising revenue; today over 80 per cent comes from circulation and user subscriptions.
Schibsted Media Group, Aftenposten’s owners, has adopted this strategy in their telemarketing and customer support departments which now see churn risk for a subscriber based on a green, yellow, or red indicator.
The first 100 days
Behavioural data offers the strongest correlation to churn probability. The Wall Street Journal uses ‘active days’ as one metric to predict churn. The more active days a subscriber has in a month, the less likely they are to churn. The previous behaviour of the user is also factored into their predictive model.
Both Aftenposten and The Wall Street Journal focus on the first 100 days of a subscriber. The Wall Street Journal’s data has found that a subscriber is four times more likely to form a new habit in their first 100 days. These habits increase engagement and consequently subscriber retention.
Another behaviour Aftenposten found that substantially increased subscriber retention were subscribers who had paid more than two bills. They found that this behaviour reduced churn risk by up to 2.5 times. This ties in with the ‘sunk costs effect’; in this case, sticking with the subscription because a certain amount of time (at least 100 days) and money (payments made on two separate occasions) have already been committed.
Mixing automation and personalisation
While data is used to reduce churn and many media organisations are increasingly using automation to do so, most still rely on telemarketing. Human interaction between the subscriber and the customer service agent still seems to be the most effective way of getting someone to change their mind about unsubscribing.
The Wall Street Journal, like many other subscriber-based news platforms, encourages this behaviour by making telephone calls mandatory. So, while signing up for a subscription can be done with a few button clicks, unsubscribing can only be done over the phone.
The call is an opportunity to get the customer to change their mind about cancelling their subscription. This is often done by highlighting what they get access to and offering details of future news coverage such as political elections or sporting events. The customer service agent, who has access to the churn score in their CRM system, can also provide an offer or reduced subscription if they sense the caller is on the brink of cancelling.
Retaining and growing the customer base
The Wall Street Journal increased its digital subscription customer base from 1.6 million subscribers in 2018 to 3.4 million in August 2023, according to Statista. Aftenposten has seen its digital subscriptions grow from 106,000 to over 165,000 in the same five-year period. These increases have been helped hugely by their carefully adjusted churn prediction models.
The Data Business is a partner of Mx3’s Collectif Network, a strategic partner community that brings together industry technology leaders and media advisors to benefit from various exclusive initiatives and to add their voices to wider industry conversations.
To learn more about Collectif and our media-tech community, contact John Schlaefli, Chief Revenue Officer, at john@mediamakersmeet.com.
