By: Katherine Ruane, Director, Mather
Background
Founded in 1949, the Frankfurter Allgemeine Zeitung (FAZ) is one of Europe’s most respected national daily newspapers.
As digital transformation accelerates across the media industry, FAZ has emerged as a pioneer—expanding its digital portfolio and investing in innovative technologies to better serve its audience. With print circulation naturally declining and the majority of subscribers now digital, FAZ’s focus on digital growth is both strategic and forward-looking.
As part of this evolution, FAZ first built its own dynamic, propensity-based paywall to strengthen subscription performance. Over time, however, the static nature of propensity models limited their ability to respond to real-time reader behavior. To push the boundaries further, FAZ partnered with Mather to implement the Sophi User Paywall Engine, leveraging real-time AI to optimize paywall decisions and drive subscription growth.
Objectives with Sophi
- Improve paywall conversions within the dynamic article segment by at least 10%, while maintaining engagement and intercept rates.
- Uphold FAZ’s editorial standards and existing business rules throughout the test.
- Ensure a seamless rollout with minimal operational lift and rapid visibility into results.
- Establish a foundation for future experimentation with paywall and reader engagement levers.
The Path to Success
FAZ and Mather collaborated closely to design a robust testing strategy, keeping all editorial and business rules intact. This created a clean baseline for measuring the impact of Sophi’s AI decisioning versus the existing approach. Mather integrated with FAZ’s data pipeline, training the model on FAZ’s audience data. The rollout was carefully managed, gradually increasing exposure to Sophi until a balanced 50/50 test was achieved.
Throughout the process, FAZ’s internal analytics team independently tracked all key metrics, while Mather provided additional reporting and insights. Regular touchpoints ensured transparency, shared learning, and alignment with FAZ’s broader digital strategy.
Impact & Results
- Stronger Conversion Performance: Sophi’s AI delivered a 65% increase in order conversion rate on articles where it made the paywall decision. The overall order conversion rate rose by 5%, with no cannibalization across channels. The proportion of paywalled articles remained stable, confirming that the uplift was due to smarter decisioning, not increased friction.
- Steady Audience Engagement: Visits per visitor, pageviews per visit, and other engagement metrics remained stable, demonstrating that adaptive AI did not disrupt the reader experience.
- Healthy Retention: Subscribers acquired via Sophi retained just as well as those in the control group, progressing through FAZ’s offer pathway at the same rate.
For FAZ, the experience and impact with Sophi reinforced AI’s role as a strategic lever, not just a technical one.
“At FAZ, we see digital innovation as essential to our mission of delivering quality journalism to a modern audience. The Sophi project exemplifies our commitment to data-driven experimentation and continuous improvement. We’re proud to lead the way in applying AI to create value for both our readers and our business.”
— Donika Lilova, Head of Subscription, Frankfurter Allgemeine Zeitung
Explore more case studies on AI-powered paywall decisioning with Sophi.
Looking Ahead
While propensity-based paywalls have taken many publishers forward, FAZ’s Sophi results demonstrate that real-time AI can elevate performance even more—unlocking gains that mature in-house models often leave untapped.
The team at FAZ is now exploring further applications of Sophi’s AI across more content, new behavioral signals, and advanced registration strategies. As FAZ continues to innovate, it remains committed to sharing insights and best practices with the wider publishing community.
With that in mind, FAZ offers the following guidance to publishers considering the role of AI in their subscription strategy:
“Start with a clear hypothesis and measurable goals. Ensure editorial standards remain central throughout experimentation. Collaborate closely across teams—editorial, data, and technology—to maximize learning and impact. Most importantly, be open to iterative testing and continuous improvement. AI is not a ‘set and forget’ solution, but a powerful tool for ongoing optimization and audience engagement.”
