I have cought up with Dr. Volker Stümpflen, Head of Data Strategy and Operations at Mediengruppe RTL, to talk about the use of real-time analytics. He is primarily responsible for the development and expansion of data science activities and the realisation and monetisation of specially analysed data models with a focus on user behaviour. He has extensive experience in large database systems and applications, based on his previous activity as Director of Data Science, with a focus on data-driven risk management in the financial investment sector. His other stations included the long-standing management of the Biological Information Systems Group at the Helmholtz Centre in München, with a focus on semantic big data systems in biomedicine. He was also founder and CEO of Clueda AG, a multi-award-winning fintech startup (including the Best in Big Data Award, winner in the Disruptive Technology Challenge Big Data category).

Has analytics reached the board level in your organisation? Is analytics part of the corporate strategy?

Absolutely! Two or three years ago it was different. Even though our core business of linear TV continues to run very well, RTL is developing into a four-screen provider across all terminals. This is still going very well, but we are preparing for the future. Data, analytics and machine learning play a much greater role here. The reason is very simple: We want to understand our users better, get closer to them. Ultimately, our aim is to offer content and market advertising opportunities to as specific target groups as possible. To this end, it is important that we gain insights and act more data-driven.

Is real-time analytics used throughout the company?

Our approach is that we want to bring analytical thinking into the entire organisation. For example, we decided on self-service BI with appropriate visualisation tools as a method. The department is able to – largely independently – find new insights and make decisions based on them.

Would you also speak of democratisation of data?

Absolutely, this is a central component of our strategy. We are implementing a cloud strategy based on this. This will allow us to define and provide a uniform data lake in the future. This makes it even easier for the departments to monitor and control their business using analytical methods.

What do you think of the idea of an “analytical platform,” and what would you understand by it?

I see two aspects. On the one hand, it is a question of implementing rather complex analyses that are rarely time-critical. We then use Hadoop-based infrastructures in batch operation. A second area, however, is the ability to perform real-time analytics using streaming technologies. Here we also have a look at tools such as Spark, Kafka or cloud-based services from Google, for example. So we already have initial approaches in operation and want to expand this further.

Technically speaking, real time is relatively new to us. We have many sources – especially our websites – which provide a lot of information in data streams. This is about merging, for example, to find the next best offer for the individual user.

What challenges do you face?

Technically, I don't see much trouble. There are enough solutions on the market that can be used here. A more important challenge is a – possibly Germany-specific – difference in how data is handled from a cultural perspective. For example, many of our conversations focus more on the limitations of data usage, while outside Germany it is much more on the possibilities of data analysis.

In our case, we are not interested in the individual user, so we have no problems with data protection. It is about anonymised profiles of user groups.

What about attracting and retaining the right talents among employees? A topic for you?

No question, we need more data scientists. We hardly find any experts on the market with the necessary knowledge in the field of big data and real-time analytics. Given the large number of our projects and plans, this is a real bottleneck. When we have them on board, we try to give these experts the exciting topics. We want to give them a deep insight into our business so they can see how their contribution is driving real change. This is only possible if we give them enough room for their own decisions at the same time.

We need more data scientists. We hardly find any experts on the market with the necessary knowledge in the field of big data analytics.

 

How do you then organise innovation? In a lab?

We are indeed considering creating our own ecosystem. Our previous “hub and spoke” concept has enabled us to bring interdisciplinary teams together in an uncomplicated manner. The fact that we always bring in two or three of our subsidiaries in a project team has proven its worth. These include technology companies, for example, but also the media group itself or the digital subsidiary. Real cooperation is really the key – also to be able to transfer a project into production afterwards.

What role does your division play in this? Do you see yourself as a central source of know-how?

We are the central point for big data, and as such are in close contact with our IT, which of course operates an appropriate infrastructure. As a “centre of excellence” for smart data, we provide data science services – also in exchange with colleagues from the entire Bertelsmann Group.

How important is the rapid implementation of findings in action?

This is becoming increasingly important to us. As soon as we have identified and analysed a user behaviour, for example, we want to give him a suitable content recommendation. This must be done in real time, if possible, while the user is still on the website. This is also interesting from an editorial point of view, for example, by analysing which topics are currently exciting and where a suitable offer can be placed.

Real-time analytics secure rapid implementation of real-time findings for media group RTL. #Analytics Click To Tweet

Finally, a question about the current hot topic, artificial intelligence.

I don't like the term; it felt a bit worn out in the ‘80s. Machine learning or cognitive computing is better and closer to what really happens and what we do. But when we are asked to talk about our projects, we also speak of “artificial intelligence.”