Leveraging Technology and Data to Tackle Emerging Risks

Dr. Christoph Nabholz spent five years working toward a forecast that Swiss Re's senior stakeholders could trust. When the breakthrough came, he says, it was not simply because the team had an impressive AI model. Bringing expert judgment into the forecast, and showing why the combined result deserved confidence, made the difference.

That story captures the central question in this April 2023 episode of Innovation Rockstars: how does research become useful when the future cannot be read directly from historical data? As Chief Research Officer and Head of Research and Engagement at the Swiss Re Institute, Nabholz connects internal expertise, academic research and client dialogue. His account offers R&D leaders a view of innovation in a business where understanding uncertainty is the work, rather than an obstacle to be eliminated before it begins.

Organize expertise around the decision

Reinsurance sits behind the insurers familiar to consumers. Nabholz explains that Swiss Re takes on risks from insurance companies and manages them across a diversified portfolio. That makes the ability to understand future risks central to its business.

The Swiss Re Institute brings research capabilities together around an aligned agenda. Nabholz describes matching internal specialists with external researchers to address knowledge gaps, improve forecasts and develop propositions with clients. Universities provide access to expertise the organization can combine with its own experience.

His working day involves considerable engagement: finding the right people, connecting their knowledge and discussing what the research means. Dialogue is also the habit he would keep if asked to choose only one. In this setting, coordination is part of producing useful research. A technically strong model still needs people who can question its assumptions and explain its implications to those making decisions.

“Foresight can only work if you have the whole spectrum of individuals that help you actually scope out what the different options are that you need to consider.”

Look at how risks interact

Nabholz describes a risk landscape shaped by connections. A pandemic, an energy crisis or a geopolitical disruption can affect supply chains and business activity through several routes. Looking at each event in isolation may miss how effects accumulate or reinforce one another.

He recalls Swiss Re's earlier pandemic modeling to illustrate another distinction: recognizing a category of risk does not tell an organization exactly when it will arrive or how severely it will affect the business. Forecasting remains necessary even where an event is conceivable.

For research teams, this changes what counts as a useful view of the future. A list of trends is a starting point. The harder work is understanding the relationships, exposures and assumptions that connect those developments to the organization. Nabholz's emphasis on diverse expertise follows from that problem. Different perspectives help reveal possibilities a single specialty may overlook.

Turn information into a reason to act sooner

The conversation moves from compensating for a loss toward helping prevent or limit one. For a reinsurer, Nabholz explains, that requires partnership with insurers that have the relationship with end customers.

Water-leak sensors provide a tangible example. The technical capability to detect a problem creates an opportunity to intervene before damage escalates. Yet a sensor only helps if someone installs it, understands its value and acts on the information. Nabholz therefore connects data with engagement and incentives, rather than treating collection alone as the solution.

He describes the possibility of sharing benefits when preventive action reduces losses. The underlying design question is whether the customer can see why participating is worthwhile. It connects a technical proposition to behavior, which is often where a promising preventive service succeeds or stalls.

These are examples of the direction he sees for innovation, not a claim that every household already has such a service or that every risk can be prevented.

Use reliable external data to remove a customer burden

Flight-delay insurance offers a different example of how data can change a process. Nabholz describes using flight information to determine a delay and trigger payment, reducing the need for the customer to submit a conventional claim.

“It's fully dynamic. And the data was there. That's why we did it.”

The appeal is concrete. Someone whose journey has been disrupted does not want another lengthy administrative task. An observable event can support a simpler service when the relevant data and partnership are in place.

For R&D leaders, the example suggests looking at friction as closely as technical novelty. The useful innovation may lie in connecting an available dataset to a decision and a customer outcome. Nabholz also identifies other claims and engagement processes where digital tools can simplify interactions, while acknowledging that fraud prevention remains part of the work.

Make the judgment understandable

Nabholz is positive about AI's contribution to modeling, but places conditions around its use. He stresses trusted underlying data, appropriate access and the ability to explain and stand behind decisions. In his account of a regulated industry, a model's output cannot be accepted simply because it is technically sophisticated.

Historical data also has limits. Expert networks help interpret what may change and how future conditions could differ from previous experience. That is the lesson behind his five-year forecasting story: the model and the human judgment needed to become convincing together.

Research leaders face a related challenge whenever they ask a business to act on an uncertain future. They need a strong analytical foundation, people willing to challenge it and a clear explanation of why the result should inform a decision. Nabholz's story gives the final step its proper weight. Building confidence in the forecast was itself a substantial research and leadership achievement.