context was our ability to test in a highly controlled, low-risk environment with clear performance metrics. This format allowed us to directly A/B test AI-generated versus human-made assets, isolating the impact of the marketing materials on engagement and conversion without compromising the artist’s integrity or user trust. We were also mindful to ensure the content was on-brand, emotionally resonant, and aligned with the artist’s identity, which involved iterative prompt tuning and creative oversight. The results validated our confidence: AI-generated assets consistently outperformed human-made content with higher click-through rates, higher engagement, and lower ad costs. Importantly, we didn’t frame AI as a marketing replacement but as a force multiplier, enabling faster iteration and greater reach while keeping human input in the loop. This hybrid approach built trust and proved that AI can enhance audience connection when deployed thoughtfully. Jochen Hartmann: We drove confidence through using rigorous methodology that combined state-of-the-art gen AI with controlled field experimentation. BMG’s partnership with TUM provided academic rigor, systematic validation, and access to state-of-the-art models, and the template- based approach ensured brand safety while enabling creative variation. We’ve seen promising signs in this area. AI-generated ads, created using existing assets such as albums and singles artwork, are often perceived as more creative than human-made alternatives. There is also no notable difference in the perceived artificiality of the ads. While these results are encouraging, they reflect creative collaboration within established artist identities rather than the creation of entirely new content. McKinsey: Where is the project anchored within BMG? Who is driving it? In your view, how important is it for such initiatives to align with a centralized AI strategy? Thomas Coesfeld: The MAGE project is championed by BMG’s corporate strategy team, which provides the strategic direction and sponsorship needed to explore AI as a growth lever. It was researched, developed, and tested in partnership with the Technical University of Munich, allowing us to tap into cutting-edge academic expertise and experiment rigorously in a structured, low-risk environment. The project was researched, validated, and implemented in close cooperation with both the marketing and catalog marketing teams, ensuring that the solution was grounded in real-world campaign needs and artist strategies from day one. After successful validation, MAGE was implemented within the marketing function, with ongoing collaboration from catalog marketing to guide asset deployment, performance feedback, and iteration. In my view, having this kind of cross-functional alignment under a centralized AI strategy is critical. It ensures scalability, ethical consistency, and operational relevance. It also allows innovation to flow across the business rather than stay siloed, which is what enabled MAGE to evolve from a pilot into a scalable capability that fits seamlessly into BMG’s broader gen AI road map. McKinsey: As a front-runner, what are the biggest lessons you’ve learned during implementation? What advice would you give to organizations just starting their journey with gen AI and agentic AI in marketing? Thomas Coesfeld: Implementing gen AI in marketing is as much about organizational adaptation as it is about technology. Organizations need to be agile and mentally flexible to continuously adapt to evolving ecosystems. From an operational standpoint, AI works best when it’s not treated as a bolt-on but as the default foundation. Start by designing an end-to-end automated flow with no human steps 46 State of Marketing Europe 2026