What data strategy is best for a subscription business?

Posted On by Dan Elman

The press has not had an easy time of it the last couple of decades. The availability of information on the internet has led readers and advertisers to migrate from print to online. But while the barrier to entry has become much lower for new publications, traditional brands like the Financial Times have succeeded in holding the fort. Not only because they lead the field in journalism. They are also top notch in database marketing, as I found out when I recently chatted with the FT’s Director of Data Strategy and Governance Mark Beckwith. He left me with a lot of food for thought.

Closer to the click

It is sometimes said that data is the new oil or the new currency. For modern marketers, it’s their lifeblood. While many of the core principles of marketing have persisted, the tools have changed. So how can we apply them?

“Targeting the right people for the right thing at the right time through the right channel has not changed at all,” says Mark. “With the shorter attention span of people, it has even become more important. The tools and technology have allowed us to become much more relevant and to react to behaviour or triggers much quicker. You could say that we have come closer to the click. We get into the consciousness of the recipient at the right time. When I started at the FT, analysing the impact of a campaign was a question of days, now our colleagues in marketing demand (and get) these results within a day.”

Healthy subscriber engagement

Data allows an online publication to segment their target audiences and adapt marketing initiatives to each segment. At the FT, segmentation is used, inter alia, to decrease churn in the subscriber base and to increase the number of new subscribers. When the FT had a ‘metered-model’, for their prospects, they were able to identify who was most likely to subscribe and the most appropriate moment to speak to them. “We created a ‘propensity to subscribe model’ using various facets of user behaviour we saw in those who had previously subscribed, that allowed us to produce a score for every user, every day, in order to identify the most likely prospects who would become a subscriber.”

We then used that to present targeted offers to those prospects that were not available elsewhere.” For existing subscribers, a similar approach is taken to detect risk of churn. “We created a segment of ‘at risk’ users to ensure that we continually reminded them of the value of their subscription and sent relevant content based on what they had previously consumed. Newsletters are a great tool to maintain that engagement.”

Predicted to disengage

Taking that model even further, the FT used data science to predict the score. “We created a segment that was called ‘Predicted to disengage’. These people were engaged but showed signs that they would decrease from their current level. This allowed us to target them and keep their engagement at a healthy level. There was also a study conducted by a sociology professor who took 40 paid media subscriptions and engaged with all of them for some time and then stopped engaging. He called that the ‘Zombie Effect’. We were very pleased to hear him say, at a conference, that the FT was the only publication that recognised this and interacted with him to activate him again.”

Sticky customers and lifetime value

Closely linked to reader engagement is lifetime value. For a publisher, it is better to have readers staying around longer than having to chase new subscribers all the time. “We want to acquire the stickiest subscribers possible. Lifetime value is a big thing. One of the things we’ve learnt is to ensure we take a holistic view of subscriber value - for instance, if we see that it would make sense from an LTV perspective, to transition B2C subscribers working in a company that has a corporate license, onto that corporate license, then we focus on achieving that outcome.

Next best offer

Data science and machine learning (ML) also go a long way in helping decide what product to offer to what prospect. This has proved quite successful at the FT. “Last year our data science team built an algorithm to understand whether we should offer an anonymous user on our website a trial offer or a full subscription. We found that we were often leaving money on the table by offering a 1 pound trial to someone who was likely to take a full subscription anyway. This was incredibly successful, but the success was only possible through a good collaboration between the data scientists who developed the algorithm and the engineers who integrated it into our front end.”

Human in the loop

As we are automating ever more, and Artificial Intelligence (AI) is getting embedded in marketing automation (think Salesforce Einstein), the question is how far will we go: will AI replace marketing technologists and data strategists one day? “I would rather have a job than a robot have my job, so I am a bit biased,” said Mark.

“At the FT we have gone from fairly simple propensity models to machine learning, the human element of that has been in pointing it in the right direction. And in making sure that automation does not jar user experience. The FT is a brand that is known as a gold standard for journalism. Our subscribers expect the same level of quality and competency from our marketing efforts. I think, at the moment, that we still need humans who understand the brand and can communicate that to our subscribers. But do ask me again in ten to fifteen years’ time.”

Last year, FT.com passed the threshold of 1 million online subscriptions. Data strategy and marketing automation played a key role in getting there. And with the further development of AI and ML, the future definitely looks bright for the publisher.

Interested in listening to my entire conversation with the FT’s Mark Beckwith? You will find the podcast here. And while you are there, don’t forget to subscribe to the ‘Dear Marketing Automation’ podcast.