Programmatic advertising continues to grow in importance for publishers, becoming a dominant share of the digital revenue mix.

But while demand surges, the ability for publishers to optimise their share of programmatic revenue remains constrained. The reality is simple: much of the control lies in the hands of large technology platforms, making it difficult for publishers to actively manage yield beyond a few levers.

One of the most impactful of these levers has traditionally been floor price optimisation. But optimising floor prices — especially manually — is far from trivial. It’s a time-consuming, complex task that requires continuous testing, access to granular data, and deep analytical capabilities.

For many news publishers, particularly when it comes to smaller brands or low-traffic properties, the ROI of manual floor price optimisation simply doesn’t add up.

At our company, we asked ourselves: What if AI could help us solve this challenge?

That question led to the development of Floorian — an AI-powered, automatic floor price manager designed to bring intelligent, dynamic pricing to all parts of our portfolio, including the long tail of smaller brands that often get overlooked.

Why AI makes the difference

What makes Floorian different is its ability to optimise floor prices continuously, using real-time data, on many data points and without manual input.

It is built to be scaled across our advertising portfolio, evaluating and adjusting the floor prices as needed — daily or even with higher frequency — based on the latest auction dynamics. This automation allows us to conduct constant testing and adaptation, which was previously impossible given the resources required for manual management.

Crucially, Floorian helps us eliminate manual work and errors from the optimisation process. By basing decisions on performance data and leveraging machine learning models, we can make faster, more accurate pricing decisions that translate into higher yield.

Furthermore, we are able to relocate the high-value human resources to generate even bigger ideas and impact for our advertising customers.

Real results, rapidly delivered

We started by piloting Floorian with a small group of brands from our media portfolio — specifically those that had previously only seen their floor prices updated monthly or even quarterly due to low priority or complexity.

Within weeks of deployment, we saw meaningful performance gains compared to our manually managed setup, leading us to expand the rollout across additional brands.

The speed at which Floorian delivered impact gave us the confidence to continue the development of Floorian. Our goal is to continue the rollout even beyond Switzerland to all our markets in 2025 and beyond.

What’s perhaps most exciting is that we’re expecting improvements not only on smaller brands but also due to the bigger data volume on bigger brands, which will have an even higher impact in the long term. Even modest improvements at scale translate into significant topline impact, making Floorian one of the major AI-driven revenue growth cases in our group.

Lessons learned and what’s next

Our journey with Floorian is a testament to what’s possible when publishers take ownership of their optimisation tools and data. Rather than accepting platform constraints as a given, we leaned into our internal expertise in data science and technology to build something purpose-fit for our needs.

As we continue to scale the tool, we’re also exploring integrations with other parts of the ad stack, such as proactive predictions of trends and fine-tuning based on market trends and seasonality.

With AI continuing to evolve in quality, performance and cost effectiveness, we see further potential to combine performance-based optimisation with creative intelligence and campaign planning.

AI alone isn’t a silver bullet nor a strategy on its own — but it can be the right tool when applied to a clearly defined challenge with the right data and team behind it, to unlock real business value.

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