How Swiss news publisher NZZ built a flexible paywall using machine learning
![agency](https://digiday.com/wp-content/uploads/sites/3/2018/01/money.jpg?w=1030&h=579&crop=1)
There’s more than one way to build a paywall. Over the last year, Swiss news publisher Neue Zürcher Zeitung has been using a payment system that is personalized to the individual based on hundreds of criteria.
NZZ requires people to register and eventually, pay. But when readers get these registration and payment messages and how those messages look varies based on predefined rules, dozens of A/B tests and machine learning.
“If we’re to be successful in paid content, we need to individualize the experience with our product and the product itself, and automate our marketing approach,” said Steven Neubauer, managing director at Neue Zürcher Zeitung. “Based on that hypothesis, this approach will increase engagement, retention and conversion rate.”
The 200-year old German-language publisher said it has increased conversion rate by fivefold in the last three years by using this system, with 2.5 percent of people who view a payment message prompted to subscribe.
NZZ now has over 150,000 paying subscribers, over half of which have digital as part of their subscription package and have accessed the site digitally in the last three months, Neubauer said. The site has nearly 600,000 registered users, an increase of more than 40 percent in 2017, and is adding between 10,000 and 12,000 per month, he said.
NZZ uses 100 to 150 rules that account for data like reading history, device and time of day to alter the messaging, text, placement and color of the pay prompt readers see. For instance, between 5 a.m. and 9 a.m. when people are commuting, they’re reading on their smartphones and won’t want to put in payment details, so no payment messages are shown during this time and people can freely access the site.
NZZ is constantly testing different combinations of these rules while machine learning identifies patterns that are fed into its algorithm. Since 2018 it has scored registered users on some 30 different attributes, like time spent on articles, frequency or how many newsletters that receive, to score them on a propensity to subscribe scale. When readers reach the top 20 percent of the propensity score they are served the payment prompt that suits their needs. For instance, heavy users who have registered but not yet subscribed tend to convert higher when they see an offer for an annual rather than a monthly subscription. NZZ has also linked article metadata with user journeys, so someone who has read a high number of articles on banks in Zurich, for instance, will receive a payment message on the next relevant article they read. Adding personalized greetings to landing pages increased conversion by 25 percent.
Rather than change the subscription product for each reader, the way that the value is communicated changes.
“We play around with the threshold,” said Neubauer, adding that people can see payment messages after they have read five, eight, 11 or 13 articles. “Ultimately, the goal is to not disrupt the product experience; we only want to disrupt when someone is willing to pay. You will never hit 100 percent success. Sometimes there is the need to indicate this is a pay product — ideally only when the lead is warm enough, willing to pay, and it’s an offer that fits the needs they are looking for.”
Advanced publishers have been using propensity modeling and machine learning to spot patterns in reader behavior to subscribe for some time, said Greg Harwood, director at strategy and marketing consultant Simon-Kucher & Partners, citing The Wall Street Journal’s ongoing work in this area.
“Market conditions can change, competitors behave irrationally, external factors come in to play, none of which can be solved for through predictive analytics,” said Harwood. “Treat it as a lever to achieve the defined strategy, rather than being the strategy itself.”
NZZ gets 60 percent of its revenue from subscriptions and less than 40 percent from advertising. Roughly 10 years ago, direct reader revenue accounted for one-third of the total. Ad revenue will further diminish, with print advertising revenue declining by double digits year over year in Switzerland, said Neubauer.
Until recently, NZZ offered subscribers two bundles: 50 Swiss francs ($50.21) a month for digital access and a PDF version of the newspaper, or 68 Swiss francs ($68.28) for digital plus print. Last November, it introduced a digital-only access offer for 20 Swiss francs ($20.10), which is proving popular, particularly with younger readers. In the U.K., a print and digital subscription to The Times of London equates to 51.37 Swiss francs a month ($51.60).
According to the Reuters Institute Digital News Report, 12 percent of people in Switzerland pay for news, less than the U.S., where 16 percent pay for online news, and more than the U.K.’s 7 percent.
More in Media
![](https://digiday.com/wp-content/uploads/sites/3/2024/06/ai-human-agreement.jpg?w=439&h=277&crop=1)
Media Briefing: The 2024 media glossary, pt. 2
In part two of the 2024 Media Glossary, we tackle many of the terms you may have heard in conversations about generative AI as well as increasing the value of publishers’ on-site content.
![](https://digiday.com/wp-content/uploads/sites/3/2024/08/Digiday_1030x600-copy-1.jpg?w=439&h=277&crop=1)
Amid layoffs and cost cutting, Time CEO Jessica Sibley is expecting a ‘very strong second half’
Time CEO Jessica Sibley joins the latest episode of the Digiday Podcast.
![](https://digiday.com/wp-content/uploads/sites/3/2024/07/tennis-money-digiday.jpg?w=439&h=277&crop=1)
How IBM and the US Open are using Watsonx to create more AI-generated tennis content
Additions include match reports delivered within minutes and AI-generated long-form articles edited by a human editorial team.