In 2025, we had the rare privilege of delivering one of the best-performing TV commercials we’ve ever produced: “You Have to Check.” The spot resonated strongly with audiences, delivering outstanding performance metrics and reinforcing the power of clear messaging paired with strong creative execution.
But success came with a challenge.
Due to image-rights limitations, we didn’t have the usage rights to feature our brand ambassador Jürgen Klopp’s image in Germany. This meant that one of the most recognizable elements of the commercial couldn’t be used in one of our most important markets. The question quickly became: how could we keep the performance of the spot intact without its star?
At the same time, this limitation opened up an opportunity. Beyond simply solving the rights issue, we are continuously aiming to broaden our reach and strengthen creative testing to drive higher traveler engagement with our brand. And this test underscores our approach, we essentially wanted to isolate the impact of the creative concept versus the use of a high‑profile brand ambassador. Specifically, we wanted to understand how much uplift was delivered in Visits per Million of Contacts, and whether the ambassador presence alone was a decisive driver of performance.
Designing a Real-World A/B Test on Live TV
The most accurate way to evaluate Klopp’s impact was to run a true A/B test in live television environments. That meant keeping everything identical—media weight, markets, timing, creative structure—with one single variable: replacing Klopp with another actor.
This approach wasn’t entirely new to us.
Back in 2018, we faced a similar challenge in the US. We had a promising spot featuring Tim Williams, but reshooting the entire commercial with different actors for multiple countries would have been prohibitively expensive. Instead, we shot the spot on a green screen with Gabby Williams and replaced the lead actor in post-production. It wasn’t glamorous, but it worked—and it allowed us to scale efficiently across markets.
Fast forward to 2025, and the landscape looked very different.
The Promise of AI—and the Reality
With the rapid acceleration of AI-powered video tools, we assumed this time the task would be significantly easier. The idea of recreating or replacing a lead actor without reshooting felt like exactly the kind of problem AI was designed to solve.
So we got to work.
We explored a range of emerging tools, including VEO 2, Runway Gen-2, and several other experimental platforms. At that point, more advanced solutions like VEO 3 or Nano Banana weren’t yet available, but expectations were still high. On paper, the promise was compelling: no physical shoot, no studio, no talent logistics—just smart prompting and clever post-production.
After several weeks of intensive testing, however, reality set in.
While some results were promising in isolation, the tools simply weren’t production-ready for what we needed. We encountered recurring issues with character consistency between shots, subtle changes in facial features, and difficulties ensuring the actor performed the exact actions required. Even small deviations broke continuity and undermined the credibility of the spot—something that becomes painfully obvious in broadcast TV.
In short, AI could get us close, but not close enough.
Back to Basics: Choosing Reliability Over Novelty
At that point, we made a pragmatic decision: abandon the AI shortcut and return to a proven method.
We chose to reshoot the commercial with a real actor on a green screen, using the same technique we had successfully employed back in 2018. It wasn’t the most exciting solution, but it was reliable—and reliability mattered more than novelty.
Because this approach required more coordination and time, the process had to be carefully planned. Our producer, Karo Güldemann, together with Matthias Cieschinger, mapped out the entire workflow, from casting and shooting to post-production integration. With the support of an external post-production house, the team ensured everything was delivered on time for the live test.
The resulting spot was aired across multiple markets, including the US, the Netherlands, and Germany—where it ran alongside other creative concepts.
Results That Justified the Decision
From a performance standpoint, the outcome was clear: the alternative version did not underperform. While it didn’t dramatically outperform the original either, it delivered solid results that validated both the creative execution and the testing methodology.
Most importantly, the test gave us something far more valuable than speculation: real-world data. We were able to better understand the incremental value of a brand ambassador’s presence, assess the strength of the underlying creative idea, and confirm that the spot itself was robust—even without its star ambassador.
In that sense, the project was a success on multiple levels.
Key Learnings from the Process
This experience reinforced two important lessons that will continue to shape how we approach innovation and execution.
First, while AI techniques are advancing at an incredible pace and often look extremely promising, they don’t always deliver at the level required for high-stakes production—at least not yet. For certain use cases, especially those demanding frame-perfect consistency and broadcast-ready quality, traditional methods remain the safer choice.
Second, the method is never the goal. Whether we use cutting-edge technology or established techniques, what truly matters is achieving the objective. In this case, that meant delivering a functional, testable commercial on time, gathering meaningful insights, and protecting performance across markets.
Innovation is essential—but knowing when not to use it is just as important.
A big thanks to our Creative Lead César Gaspar for sharing these insights!