On Wednesday, INMA members were treated to a live Webinar exploring the business applications of data at Swiss media company NZZ (Neue Zürcher Zeitung) and its push for personalisation. 

NZZ, founded in 1780, is a small publisher in Switzerland with a legacy and strong brand that the organisation is always aware of.

Daniel Ammann, head of portfolio management at NZZ, told the INMA audience that 53% of NZZ’s US$152.6 million revenue comes from reader revenue. “When we talk about the digital subscription business, we see it as we do many other businesses: a conversion funnel of we reach people, engage them, then sell them something — and, of course, how we increase them in our product portfolio.”

Ammann led INMA members through each of these four steps:

• Reach: Very early in the process, NZZ tries to get readers to register and thus become identified users. The company currently gets 10,000 registrations per month in a limited market and small population.
• Engagement: Personalised content recommendations.
• Sales: Convert readers into paying users. The news company has seen a conversion rate at the paygate of five times over the last five years.
• Increase: News brands must have the right product portfolio for their customers. NZZ saw a 60% year-over-year increase in digital subscribers in 2018.

Identifying users

“We introduce our reg-gate very early on in the customer journey,” Ammann explained. But it’s not as simple as just having a registration gate. News media companies must actually give users something that will entice them to register.

“The most obvious thing to do on a news Web site is content. As an anonymous user, you can only read five pieces of our content per month. And after that you have to register,” Ammann said.

A second way NZZ triggers registrations is with newsletters, which require registration. With almost 30 newsletters currently in its arsenal, NZZ has seen success with them. “You really cannot have too many newsletters because it’s a good tool to let the users read what they are interested in and be part of your community.”

The third tool used to trigger registrations is features. A user must be registered to access all the features offered by NZZ. Without being registered, the user does not have the same experience.  This includes tools such as a reading list, “My NZZ” and other personalisation.

“The user experience as an anonymous user is very limited, and this is on purpose as the prompt to register and get the full, enhanced user experience,” Ammann said.

Personalising content

Once a user is registered, NZZ uses data to start tracking the user to deliver a personalised experience. “We try to create a smarter cross-channel news experience for our readers,” Ammann said. “The goal here is really to make the experience on our platform more relevant for every user.”

As part of that, My NZZ offers the user a personalised article list as well as “what happened since you last visited” features. Smart article teasers, based on the user’s behaviour, and a personalised newsletter are other tools. This personalised content feed is created with an algorithmic curation method.

“This consists of three elements. The first is an editorial score,” Ammann explained. “Then we also add a crowd score, which articles are the most read, the most liked, the most shared.”

Finally, and most importantly, the personal score is then added, which consists of the user’s reading history and topics of interest, while ignoring articles previously read, Ammann said. “Out of this feed, we then create data products.”

Some of the things that the NZZ team has learned from this personalisation initiative include:

• 20% of subscribers use My NZZ.
• 84% see it as an added value.
• 16,000 newsletter subscribers were attained within the past few months.

“Although the subscribers like it, they don’t use it as much as we thought they would,” Ammann said. “For it to be a really powerful, engaging tool, the traffic is not really high enough.” Wording and integration are key to this, as is making sure that different feeds fit different purposes. “Explain to the users what they will get when they use this product.”

Increasing paygate conversion

NZZ introduced its paywall, which was a classic metered model, in 2012. “It was a bold move at that time because it was the first paywall introduced in Switzerland,” Ammann said.

The conversion rate was low, however, and because ownership was placed within the IT department, the rest of the organisation didn’t have access to it. That changed in 2014 when NZZ developed its Flexible Rules Engine.

“We introduced nine dimensions on which we can adapt the paywall, make tests, and improve it,” Ammann said. “Organisationally, it was moved to the business unit who was now able to work with it.”

One example of a change made at that time was adding a personalised greeting on the Web site for each logged-in user, which improved conversion rates by 25%.

In 2017, NZZ went one step further with its paywall model to develop a dynamic paygate and really start A/B testing this with target groups. “We really used data mining to recognise patterns in the behaviour of our users,” Ammann added.

Ammann used an example of how the new, dynamic paywall worked with pattern recognition for pricing preference. “Not every user has the same reaction to prices. Some prefer lower prices, but some readers actually prefer high prices. Different prices work differently with different target groups.”

The last step in the paywall journey has been the latest iteration in 2018, with the dynamic paygate v1.0, using propensity scoring with machine learning. “We have a propensity score for each individual user on NZZ,” Ammann said. “More than 400 factors go into that score. The most important ones are time since registration, time since last visit, number of devices used, etc.”

It’s not a real-time score yet but is generated overnight. One example of how they used this propensity scoring is for people not in the top 20%, the standard rule set would apply. Those users who were in the top 20% score of likeliness to buy would be moved into the A/B test.

“They either got the standard rule set or a specialised rule set with a direct prompt.” This showed the paywall no matter how many articles the user read. “This drove up conversion rates by 82%,” Ammann shared.

Adapting the business model

When NZZ started improving its registration and paywall models, the team realised it was now reaching the right people — but perhaps it did not have the best product portfolio for those people.

“We realised that to actually reach new target groups and to make our focus on the reader market and subscription business means that we must give products that they like,” Ammann said.

This included a digital-only product, which was also much less expensive to produce. “It worked very well. Cannibalisation was very low, and virtually no one downgraded. We really saw that we could attract new readers to the portfolio.”

NZZ also started enriching its product portfolio geared toward its customer needs. This included a new student edition, reaching the 33% of new subscribers who were students. The team also realised the German market, in addition to the German-speaking Swiss market, should be targeted. It created a new product specifically geared toward Germans, resulting in 14,000 new subscribers in that country. Another new product is geared toward NZZ’s corporate customers. In this platform, corporate customers can manage their subscriptions very easily.

Results

With these changes, digital subscriptions increased dramatically for NZZ. “We added 60% more digital subscribers over the past year,” Ammann said. Total subscribers also went up to counter declining print subscriptions by more than 4% to a total of 159,000 subscribers in July 2019.

“We need many more subscribers,” Ammann said. “Our new goal is 200,000 subscribers, and this has to continue for the next few years to reach our goal by the end of 2022.”

Ammann concluded with sharing key takeaways presented in the Webinar: “We took the risk with our product portfolio, and I think that’s important if you want to succeed with digital subscriptions.”

Q&A

INMA: Which initiatives specifically do you use to get registered users to become paying users?

Ammann: We try to engage them with personalised content, and then we try to get them to come into our paygate that is also personalised. It’s not that there’s one specific initiative works for all users; we see that different things work for different target groups. We really rely on our data and algorithm to help us determine what to use.

INMA: Do you offer discounts or special offers to go with registration?

Ammann: No, not with registration. You get more content and the newsletters and so on. But we do offer discounts on the paid subscriptions. What we’ve found is that the monthly subscription people like very much, so we often include three months. But you only have to pay for one, and then after that it goes monthly.

INMA: Which strategies have you found as the most successful to explain these personalisation features to users?

Ammann: It was one of the learnings that we had that was very important. I’m probably not the best person in-house to answer that. We do not use the wording “personalisation” actively with our users. We use words more like “this is just for you.” We also explain what we are doing, we always have a link to an article that explains what is behind our recommendations. We’ve learned that readers appreciate this to understand how personalisation is important for them. Integrating this into the core product is important and works well.

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