Acquiring new subscribers is an expensive business. This is all the more reason to focus on keeping the existing ones content and loyal. This seemingly simple observation is the key reason behind the increased focus on customer retention as a part of the news media business model.

Classic churn analysis work is primarily concerned with detecting danger signs in a customer’s behaviour to find out what measures most effectively prevent someone from cancelling his or her subscription.

Companies usually follow up with new subscribers with a common communication flow to guide them through the first or second payment periods. However, their subsequent behaviour regarding payment methods, subscription type, or digital usage may differ so substantially that each reader has to be treated individually in order to detect behaviour consistent with churn.

A prediction model’s prime objective is to generate scores based on consumer behaviour that enables us to differentiate between various customers, and, if possible, provide prescriptive advice on what to communicate to customers.

One example of such prescriptive advice is to find all the active customers with a score indicating they are in danger of churning and whose high score is caused by low, no, or decreasing digital usage. The model may then identify one of these factors as the primary cause for concern and enable us to target the specific customers with this pattern of conduct.

Tests carried out found the retention rate in the specified segments were 10 percentage points higher among those who had received the relevant communication compared to the control group.

To avoid spamming the customer’s e-mail inbox, it is most optimal to construct relevant segments for each type of market communication. Churn scores and the identification of churn drivers is an important tool for the CRM team when working on composing the right message to the right segments rather than delivering every message to the entire customer base.

A new way of using churn score

The use of churn score to predict churn is a well-established routine in the way Schibsted works with retention. However, communicating with subscribers through e-mail and in the products still doesn’t enable us to communicate with everyone at risk of cancelling their subscription.

Schibsted’s customer service department receives more than 700,000 inquiries yearly. The department is also the most flexible in customising its messages for the needs of each subscriber.

As we experience positive results from using churn score in our retention communication through e-mail and in our digital products, the hypothesis is that a churn score prediction visible to the customer consultants would be helpful in customising their dialogue with each inquiry.

Customer consultant tools with a structured communication range

A score enables the customer consultant to engage with the subscribers as the score is displayed on the operator’s screen along with other customer information indicating whether there is an imminent danger of subscription cancellation. A green customer ought to be safe and a yellow one is a question mark, while a red customer is a definite danger case. This provides the consultant with valuable information and guidance on how to approach each customer to build loyalty and prevent churn.

The indication for a green light is as follows:

• The subscriber profile shows these subscribers are logged in, using digital products, and probably have no unpaid invoices. They are most likely utilising an automatic payment method and have active digital access.
• The customer consultant action shows the subscriber is most likely open to for an upsell.

The indication for a yellow light is as follows: 

• The subscriber profile shows these subscribers may not have used digital services or are not digitally active. They may not have paid by a due date one or several times. They may not use an automatic payment method but are paying by paper invoice.
• The customer consultant action is to motivate subscribers to use the whole range of the product they subscribe to. They should also motivate subscribers to use digital or automatic payment methods.

The indication for a red light is as follows:

• The subscriber profile shows these subscribers do not use digital products and may not have a digital log-in. They may have paid late one or several times. These subscribers also may not have been active for more than three payment terms.
• The customer consultant action is to provide a great customer experience and pay close attention to customer communication. They should motivate subscribers to use the whole range of the product they subscribe to and motivate subscribers to use digital or automatic payment methods.

Providing valuable data to optimise churn prevention in other channels

Using churn score continuously in the customer centre and tracking the actions of the customer consultant also provides valuable insight and data that can be used as an input in subsequent profitability analyses as well as for more generic segmentation purposes.

The churn score is also a valuable and easily comprehensible tool in the CRM department’s toolbox that, if used efficiently, may contribute significantly to decrease churn by identifying disaffected customers, understanding why they are not content, and providing recommendations on how to increase their level of satisfaction.

By challenging the use of data and insight to personalise communication, our goal is to optimise retention flows. The hypothesis is that the combination of successful sales activities and the use of a churn prediction score to optimise the retention activities will increase the length of the subscriber relationship.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .