top of page

Corporate insurance domain left behind by Data Science


The use of data science and Artificial Intelligence (AI) is accelerating in the private insurance domain within the Dutch market. IG&H conducted interviews among six Dutch insurance companies and learned the investment in people and tooling and successful use cases are piling up in the private domain. In the corporate domain things are much more basic. This shouldn't be the case…


Corporate domain is way behind

The last couple of years Data Science really took off in the insurance market. Especially in the Front Office, with applications in Commercial Pricing, Next-best-actions and personalisation of the online user experience. Investments were made to define use cases, disclose data, test and scale. And above all, a lot was learned; how to find the right (scalable) use cases? What works and what does not work to deploy models into production? How to attract the right talent? How do we organise ourselves for success? The focus has been on the private domain and not on the corporate for several reasons; The private domain has larger volumes of customers and interactions; the data is more straightforward and there are more direct channels. To start, this is a logical choice. But this has caused an unintended imbalance. A round among large all-round insurers taught us that on average 70-80% of all data science effort is spend in the private domain (and within this, mostly on property). When we compare where Gross written premium (GWP) and profits come from this is almost a mirror image: 60-65% of GWP and 80% of profits* are from the corporate insurance products.

Dogmas to break through

Insurers and intermediaries are increasingly aware of this imbalance, but this is not yet resulting in larges shifts. For this to happen a number of dogma’s need to be disproven. We outline the five most important ones here:

In practice we clearly see that many expert judgments and assessments can be greatly enhanced, or replaced, by data science techniques. Also, the more strategic and tactical decisions, like segmentation and portfolio management uniquely benefit from a more data-driven, model-based approach.


Data science methods and tooling has progressed, and we have learned how to be successful despite lesser data availability, or lower volumes. On top of this, it is a myth that AI models are by definition black boxes. We have several techniques in our tool kit to expose the inner workings and explain the results of many different powerful algorithms.


“Especially disproving these dogmas can accelerate things. The combination of business hypotheses and data science can generate lots of value in the corporate domain. Compared to the private domain there is so much we can gain here.” Peter Zwikker, Manager Data & Analytics Centraal Beheer Bedrijven

Proven value

The IG&H Data & Analytics team has experienced with clients how AI can bring enormous performance improvements in complex, corporate domains. Here below is one of the examples.


A market leader in corporate credit wanted to make it credit processes more efficient, agile and more consistent. This incumbent also wanted to be more ready to compete with the new, fintech competitors. We used AI to implement a new decision model that reduced manual expert judgments in the risk review process by ~80%. The model has learned from examples provided by experts to assess the need for a manual review. A low score means the case will no longer be looked at by specialists.


The lack of historical data was overcome through pragmatic workarounds. By using some new techniques, the model is very transparent. On top of this, specialists and managers can monitor their portfolio more frequently and consume many new insights in an interactive dashboard.


This model is the first of several we are building, and this one model already frees up ~6 FTE of specialist capacity. This capacity can be allocated to new business and to support clients with more complex risks. The estimated return on investment of this first project to be within 2 years is also very agreeable.


Next to the direct application of AI within a business process, data science also makes many other ways to improve performance possible. For example, using data science techniques the business processes can be analysed to reveal bottlenecks and inefficiencies. We recently analysed a business process at a financial services provider and used the results to formulate new business rules that eliminated 90% of manual approval steps. We are also putting big data technology to use to optimise tactical and strategic choices in portfolio management.


“We apply Analytics & Data Science in our corporate insurance domain very successfully for years. Using process analytics, we constantly optimise our internal processes and customer journeys. These optimisations lead to increased customer satisfaction and efficiency. A genuine win-win situation.” Marijn Janssen, Head of Analytics & Data Science Aegon NL

Looking forward

It is time for a radical turnaround and scaling up of data sciences efforts in the corporate domain. The results are there, and data science will play an important role in the future competition. There is serious potential to be gained and cross-pollinating with lessons learned from the private domain will help companies to accelerate in the corporate domain.


Contact

Mando Rotman

Senior Manager Data & Analytics IG&H


Jan-Pieter van der Helm

Director Financial Services IG&H


*) excluding revenue from closed book Life in private


 

A takeover wave amongst Dutch insurance brokers is taking place

Private equity is growing its footprint in the Dutch insurance brokerage industry. An attractive financial profile, large consolidation potential and the possibility for ‘multiple-arbitrage’ led them to be responsible for 21% of all acquired revenue over the last two years. IG&H and PhiDelphi Corporate Finance jointly investigate the drivers for and impact of the M&A wave that is currently taking place.


Коментарі


bottom of page