Blog | FELD M

Interview with Tim Stahringer, Program Manager @ HUK-COBURG | FELD M

Written by Eilish Prioul | Nov 19, 2026, 10:49:23 AM


FELD M in Conversation is a brand-new series of interviews featuring interesting people from the data and analytics industry. It’s essentially a fireside chat (delivered to your inbox, if you sign up here). A space for people to share their experiences, approaches, and personal perspectives – going beyond frameworks and theories.

We hope that you, The Data Collective, will find voices that resonate and inspire.

 

 

From speed boat to oil tanker, always keeping the vision in sight


“People are the center of it all, and you need to get them on board. Business is all about people. You can set up all kinds of great tracking systems and build data warehouses. But in the end, if it doesn’t make a difference to people, it was all for nothing. And every time you change something, you need to bring people along with you.”

 

Tim Stahringer | Program Manager | HUK- COBURG

It's no secret that the data world is full of pretty cool people. But that doesn’t make it any easier to decide who to interview next for FELD M in Conversation – in fact, we felt spoiled for choice!


That’s why on her last visit to the FELD M office, Eilish turned to her colleague Kai Brackmann for help: When he thinks of interesting people from the industry, who’s the first person to pop into his head? Maybe one of our FELD M customers?
Kai was quick to suggest Tim Stahringer, Program Manager of the Data Analytics Program at HUK-COBURG, which, with its 13 million customers, is one of Germany’s largest insurers for private households. He’s been working in the insurance industry for eight years.

 

An email and two weeks later – finding a time to meet is always a challenge, but more on that later – Eilish and Tim settled into their Teams call, and what began as a quick “hello” turned into two hours of chatting.

 

It quickly became clear that Tim has his own unique perspective on his work. It had all started with the simple desire to “do something to do with data”, at a time before the term "data science” even existed. Over the years, he added conversion optimization and analytics experience to his toolbelt, along with a whole host of tools and processes. But his most important lesson remains the same to this day: Always keep your vision in sight. 


Starting at the beginning: Can you share a little about your career path at the HUK-COBURG Group? 

“I was initially hired as a Customer Data Analyst at HUK24, but soon transitioned into the role of Product Owner for data-related topics.  

From there, I went from one PO role to the next. Then the Data Analytics Program was launched, where, as Product Owner, I oversaw a specific focus area at HUK24, specifically: how to integrate data analytics and marketing. And a year and a half ago, I moved to the Data Analytics department as a Program Manager.

We now have about 30 product owners and project managers in the program, all of whom work on data and analytics topics, and I support them with training, workshops, and change management.”


HUK-COBURG, HUK24… if you’re feeling lost, here’s a little context:


HUK24 is the online insurer under the HUK-COBURG group's umbrella; while the group was founded in 1933, HUK24 emerged in 2000, right in the middle of the dot-com bubble. Launching a digital insurer at that moment was a forward-looking move. Today, it remains a lean organization with around 40 employees plus roughly 200 in support, which, with the group’s backing, has developed its own unique pace. 

 

You've worked for both HUK-COBURG and HUK24. How would you characterize the two companies? 

“For me, the biggest differences were the pace and the proximity to decision makers. HUK24 was deliberately kept small. Core functions like accounting were outsourced to the corporate group level, but all strategically relevant topics were handled within HUK24. That of course had a huge influence on the work. Everyone took ownership of their area. With no management layer between you and the board, you had the freedom to think like an entrepreneur, make decisions, and put them into practice. You could think of HUK24 as a speedboat. 

By comparison, HUK-COBURG is more like a giant oil tanker. It’s stable and powerful, but it moves more slowly. So of course there is a difference: you're coordinating with 20 people, not two.

What I do have to say, though – and it's something I learned back at HUK24: If I want to get something done quickly, I do it alone. If I want to do it well, I do it with a lot of people.”

 

So what did that mean for you in practice? What was similar in your roles, and what differed?

“My approach was actually very similar: always starting with the vision. Where do I want to go? What’s the big-picture goal? 

Once you have a vision – ideally developed together with the team – you can work backward to figure out how to reach it. What are the next steps, and what ripple effects will they have? What tech stack do you need? What people and roles do you need? Wherever I’ve worked, whatever my role, that approach has stayed much the same.

There are a few differences, though. First off, my roles at HUK24 were often very focused on execution – a strong “bias to action,” as they say; getting stuff done. 

HUK-COBURG calls for significantly more planning.

In terms of how data projects themselves differed: At HUK24, if I changed something in the process using data analytics, the process itself changed. So I needed to make sure that people actually used the new process. In a small organization, that’s easy. You talk to three people, explain how it works, and that’s it. At HUK-COBURG, by comparison, we introduced a new process that affected about 1,000 claims adjusters. At that scale, when it comes to communicating that change, integrating it into everyday work, gathering feedback...you really have to get people on board. It's the only way the new process can create real value.”

 

At a time when constantly changing jobs is almost seen as a recipe for career success, you've remained loyal to HUK-COBURG for many years. What motivates you to stay? 

“It’s mostly thanks to the variety of exciting projects I’ve been able to work on. I once attended a lecture on insurance, and the professor told us: “Insurance is actually a really interesting industry, because people always think that it’s totally boring – the trick is projects. When you work on projects, you realize it pretty quickly –  there are always new developments, new technologies, regulatory changes, and work that spans departmental boundaries.” 

“What really surprised me, though, is that even as a regular employee in a large organization, you can find a few people who share your interests and work on a project together. And suddenly, you can make a difference at HUK-COBURG, build something new or introduce new processes. That’s exactly what I’ve experienced in smaller companies: Anyone can improve and optimize things. And it's just as possible in a large company like HUK-COBURG.”

 

And beyond projects, you’re also working with lots of different types of data. Can you share a few examples? 

“The insurance industry encompasses so many different professions. There are even art historians who work in the industry, estimating the value of paintings, for example. So you can imagine that the variety of data we work with is just as broad.

HUK-COBURG primarily handles personal insurance. So that covers things like insuring your car, your house, your apartment, etc., as well as legal protection and personal insurance products. In other words, everything related to the individual, primarily health and life insurance. As a result, we work with a huge variety of data, covering everything from hospital bills to photos of broken bikes, all the way to stock prices and court transcripts.

From a data protection perspective, the data is strictly separated – even across separate companies. What that means is that I can’t go and say: "This customer has insured their bike. When was the last time they were sick?" In Germany it’s really strictly regulated, and HUK-COBURG goes even further when it comes to data protection. We can only request the information we truly need.”

 

And what are some ways you use the data? 

“One example from my colleagues’ work that I find particularly interesting is deciding when to send invoices. HUK-COBURG sends out about 14 million car insurance invoices a year, and customers often call with questions about their invoices. That means we have to figure out the best time to send out the invoices so that the call center doesn't get inundated with calls.

For years, our colleagues have been developing a model to determine the best time to send an invoice to each individual customer, taking into account things like churn and call center capacities. It directly translates into measurable added value for HUK-COBURG.

Then there’s the topic of pricing. How much should an insurance policy cost? I don’t think there’s any other industry that devotes as much time to pricing as insurance does. With 14 million car insurance contracts, pricing each of those just one Euro more accurately adds up to a lot of money.

From a methodological standpoint, pricing is based on a forecasting model. The goal is to predict the claims people might make, ideally on the most granular level possible. The better the actuaries (insurance mathematicians) are at predicting reality, the better the risk selection becomes.

That means that as an insurance company, I want a group that causes, on average, €500 in claims per year to pay roughly that amount for their insurance. If that group only pays €400, then other customers have to carry the cost. If we ask them to pay €600, they might switch to another insurer. Precision pays off, and when you look at that on the scale of 14 million car insurance policies, you’re easily looking at billions in losses if you get the pricing wrong – or significant profits. We pass these profits back to our customers.”

 

That all sounds like a great starting point – plenty of data, accumulated over the years, and methods for deriving real value from it. I’d be interested to know what the other side looks like: what challenges do you face when working on data projects in the insurance industry?

“My top takeaway is that the difficult part of data projects is neither the methods nor the data. If I go through the stack - data storage, ETL, modeling - it all works smoothly. We have really talented colleagues who work exclusively on these topics.

One initial hurdle is better integration into the existing infrastructure. There are various solutions, but we haven’t quite found the right one just yet. HUK-COBURG has a complex IT infrastructure spanning many services, many developers, and of course regulatory requirements, too. A data scientist’s dream scenario is to be able to press a button and just like that, their new model goes live. But that’s not the reality. 

I think the biggest challenge in all the data analytics projects I've ever worked on is a different one, though: How do you make new tools and processes a natural part of people’s everyday work? Let’s take the example of fraud detection. We have many fraud investigators. If a data scientist wants to flag certain cases to collect data, the fraud investigator sees the cases and thinks: “Why has this been flagged? Is the system broken?” That’s why claims adjusters – the users – need to be integrated into the process. It’s fairly easy when you have 10 employees, but how do you do that when 2,000 people are affected? It’s a significant challenge.”

 

So how do you approach it?  

“What tends to work quite well is to involve those affected early on, ideally taking a co-creation approach. We want to change something, so we speak to the people who would be affected. The motto is: Let’s discuss that change together, from the very beginning, and ideally figure out together what it should involve. Those conversations can lead to surprisingly simple solutions we might otherwise have missed.

Something I thought was brilliant: In the claims processing department, we automated some of the processing steps using weather data and things like that. But when we looked at the process metrics, we realized that the number of processing steps for the automatically verified cases was no lower than for the manually verified cases. 

Something was off. So we spoke to some experienced claims adjusters, who told us that the staff were double-checking everything manually because they didn't trust the automated results. 

The solution was simple, but smart. We attached a PDF to each claim that said, “This has been checked, ✅. This has been checked, ✅. You still need to check this.” This tiny change drastically reduced the number of manual processing steps. The claims adjusters could see what had already been checked, knew they didn’t need to worry about it anymore, and could just focus on the remaining steps.”

After a good hour of conversation, Eilish’s list of questions is starting to fill up with checkmarks, too. One point that came up again and again, however, deserves a moment of its own: 

 

It just goes to show once again how important people are when it comes to this topic. As you said – the challenge isn't the data or the technology; it's always about the people.

“Exactly. A data scientist can develop a model to check whether weather data matches the claim in a couple of weeks. But getting thousands of claims adjusters to trust the results and incorporate them into their work, that’s the big challenge in cases like these.”

 

In your current role, you work with a wide variety of people. Is there a lot of politics involved, or is it more about building trust? Do you have to find “data champions” – people who are genuinely interested in the topic?

“The goal of the Data Analytics program is to make HUK-COBURG a data-driven company. That’s the vision. 

So how exactly do we do that? We have focus areas led by the various departments, each with a project manager, plus product owners who oversee individual use cases. I primarily work with product owners and project managers who take on the role of champions or change agents. We start by assessing the situation as a change team: Where do we want to go? Where are we right now? What are the barriers? Then we empower each individual to drive this change forward.

The goal is transformation: Changing processes, empowering staff with new skills, tools, approaches, and methods. When they face challenges that are too complex or politically tricky for them to handle on their own, we provide support – usually through workshops or coaching – to help them, for example, resolve conflicts, gain clarity, or develop a vision. Above all though, we champion a co-creation approach.”

 

With all the change you’ve helped bring about, your own role has also evolved over the years, moving away from working with individual claims adjusters toward a more strategic approach. How do you stay in touch with whether the initiatives you launch are actually resonating with people?

“When it comes to many HR-related topics, emotions and intuition seem to play a particularly big role. Personally, I prefer to take a data-driven approach when deciding which direction to take.

For me, there are two pillars:

  • One is the topic of enablement. There’s a lot of literature out there, the classic example being Kirkpatrick’s Four Levels of Training Evaluation, which outlines how to measure the success of training. But not many organizations actually do it. Many of them measure, say, how satisfied someone was with the training, or conduct a survey to see how much people remember, but that’s it. But what’s really exciting is the transfer of knowledge: Does the behavior carry over into everyday work?

  • The second pillar is change. How do you measure where you are on the journey of change? There’s a lot of research on this, yet very few organizations measure it, which I find baffling. For example, there’s the COM-B model, which is scientifically well-founded and describes how to bring about behavioral change. The Theoretical Domains Framework (TDF) goes even deeper in terms of scientific detail, while the Behavior Change Wheel makes it more accessible and easier to understand. 

My goal was to build simple and practical tools based on these fundamentals, and we now have a variety of these tools in use. Some of them focus on the measures: How effective were they, really? Was the effort well-spent? What were the lessons learned? 

Other tools help us see where the change stands and decide what to do next: Where will my time and resources make the most difference? Do people understand why we’re making the change? Or do they understand it in principle but lack what they need to put it into practice?

Change is a topic that often feels neglected in Germany. The methods exist, but they aren’t always put into practice. It feels a bit like the early days of data analytics: The data was there, but no one really used it. I’m doing things differently as part of our change team in the Data Analytics Program.”

 

That’s a very promising outlook for the industry’s development. At a time when so much is being disrupted by technological changes, it seems increasingly important to keep an eye on how people cope with change. Speaking of technological changes, which current developments do you view critically, and which positively? 

“I’m struggling with the whole “AI as a savior” narrative that’s being pushed really hard at the moment. Someone smart (Roy Amara) once said that change is overestimated in the short term and underestimated in the long term. We’ve all heard the huge PR hype that says we won’t need developers a year from now. I think that’s exaggerated. The much more interesting question is what things will look like in ten years. 

What I find positive is people's willingness to try things out. Give it a go, see what works. Something else I see critically is this MIT study that says that 95% of GenAI use cases have failed: 

 

Despite $30–40 billion in enterprise investment into GenAI, this report uncovers a surprising result in that 95% of organizations are getting zero return. The outcomes are so starkly divided across both buyers (enterprises, mid-market, SMBs) and builders (startups, vendors, consultancies) that we call it the GenAI Divide. Just 5% of integrated AI pilots are extracting millions in value, while the vast majority remain stuck with no measurable P&L impact. 

 

Source: The GenAI Divide: State of AI in Business 2025, a report from MIT’s NANDA Initiative, readable in full here.

 

“In my opinion, AI skeptics are lending this study too much credence. Yes, much of AI is still in its infancy, but it always takes time to figure out how to use new technology effectively.

If you use the models and tools and examine them critically, you can clearly see what’s working and what isn’t. You’re probably familiar with the Gartner Hype Cycle. I think we’re past the peak of the hype, but it’ll still be a while before the new technologies become productive and part of our everyday lives. But the claim that there won’t be any developers six months from now? No. It won’t happen that fast.”

 

Eilish: It reminds me of what you said about the early days of analytics. At first, there was hardly any demand, and then everything exploded because suddenly everyone wanted to hire data people. That’s exactly how it feels with AI right now. At FELD M, we’re seeing a lot of demand for topics like AI strategy and AI use case design, but also that many companies are still in the very early stages. To bring these topics to fruition, it seems like the topic of cultural fit is just as key. On that note: 

 

Have you ever worked on a project where the technology delivered, but the fit with the organization’s culture proved more challenging?  

“As always, it comes down to people. As an anonymous example, we had a new tool that would’ve been a huge help to the staff. It was based on various models, and it basically showed you at a glance something that it used to take a long time to find. It took maybe three weeks to develop it: data scientists and machine learning engineers can build something quickly and create value right away.

The project didn’t fail; it’s still running. The challenge was integrating it into the existing system. If a team quickly builds something in Python and then hands it over to developers who work in Java, their concerns are understandable: “I didn’t write this code, and it’s not the language we use.”

It comes back to what I said earlier: You can build something quickly and get it up and running fast. But for it to work really well and be sustainable, it’s better to involve everyone from the very beginning. That also includes taking any reservations seriously. Pushback is often dismissed as “they just don’t want to change,” but there are almost always very legitimate reasons why someone might not want to go along with something.

That's why in every project, I’ve started to ask, "What are your concerns?" To be completely honest, I don’t think I’ve ever seen someone resist something without having reason to. So if you meet pushback, you should see it not as opposition to what you’re doing, but rather as concerns that need to be identified and discussed.”

Eilish: From a marketing perspective, I can definitely relate to that. When I started at FELD M, we relaunched our website, and there were concerns about licensing expensive tools that wouldn’t get used, because that was something people had seen in the past. In my interview with Jesse Rothenberg, the Head of Data and Insights at Ableton, I learned a great term for this: organizational trauma. We all carry it around with us. 

 

Tim and Eilish are an hour and a half into their chat. When asked whether he still has energy left for a couple more questions, Tim says, “I can talk. Now there's something I can do.” 

 

I have a couple of more personal questions for you around motivation and recommendations. Then we’ll both go grab ourselves a vat of coffee! Are there any topics you resist delegating because you just really enjoy working on them yourself?

“No, not really. I’m a huge fan of Peter Drucker’s "The Effective Executive". It’s my number-one book recommendation for any manager. It’s an old one, from 1966. On the topic of delegation, it says this: For every task that lands on your desk, you should always ask yourself, “Can someone else do this better than I can?” If so, then that person should do it. So even if it’s something I really enjoy, if someone else is better at it, they get the task.”  

 

That sounds pretty fair! On the flip side, let’s say you could fully automate something, would you? What would it be?  

Tim: Finding meeting slots. 100 percent.


Eilish: Oh, absolutely. You've hit the nail on the head there.


Tim: Looking for meeting slots is a nightmare, especially with external people where you don’t have access to their calendar. Or those situations where there are three of you and none of you have time, but if you moved that one meeting, a slot would free up for all three of you. If someone wants to automate that, count me in!”

You mentioned Peter Drucker just now. What are your go-to sources for learning and professional development?

“I’m a big fan of the classics. I read books and work my way through them. With all this flood of news – LinkedIn here, the latest thing there – so much of it will be tomorrow's fish and chip paper.

My approach is based on the Lindy Effect that Nassim Taleb describes in “Antifragile”. It says that things that have existed for a long time will probably continue to exist for a long time. If we’ve cooked for the last 100 years, there’s a decent chance we’ll still cook a hundred years from now. If we’ve used LLMs for the past year, I’m not so sure whether we’ll still be using them a hundred years from now. And I feel the same when it comes to knowledge. 

In another book of his, “The Black Swan”, Taleb also shared something I’ve really taken to heart. He asks, “Do you read the news every day?” Most people say yes. Then he asks, “What were yesterday’s headlines?” And most of them have no idea. Ever since I read that, I’ve tried to avoid always jumping to the next thing. It doesn’t matter, and it ultimately just leaves you feeling anxious and worried. Everything that’s really important will find its way to you anyway.”

 

And one last question: If someone only reads the first two minutes of this interview, what's one thing you'd want them to understand about working with data? 

“People are the center of it all, and you need to get them on board. Business is all about people. You can set up all kinds of great tracking systems and build data warehouses. But in the end, if it doesn’t make a difference to people, it was all for nothing.

And every time you change something, you need to bring people along with you. That includes listening to, understanding, and discussing their concerns. Help them to understand how they can use something and how it will be implemented.

That's my summary of the entire interview: people, people, people.”

Eilish: Funny that we should end up here. My first interview with Jesse took a similar path – started with the facts, ended with philosophy. In the end, these questions are really just about people. That was also my thinking behind this interview series. In an age when so much content is generated by AI, I think it’s important to talk person-to-person. We all forget that sometimes, don’t we?  

Tim: As silly as it sounds, sometimes I think we make everything far too complicated. We’re all just people. And we often forget that. I’m including myself in that. And what you’re saying – “talking person-to-person” – really resonates with me. I can easily picture people 15 years from now saying: “I’m consciously choosing to be human.”  

 

Thank you very much, Tim, for this illuminating and entertaining interview. We hope you enjoyed reading!