The four Business Models that can guide you to answering the questions posed by AI in insurance
A well-known professor was presenting new improved automated vehicles including cars and motorbikes. Each statement he made about advances in effectiveness and safety were supported by detailed diagrams and algorithms. It was all very clear, and the audience of mostly engineers and computer scientists seemed convinced and satisfied. It was iterative improvement from sound principles.
Then after around half an hour he made a controversial statement that drew gasps from the audience ‘…you do not need AI to make self-driving cars!’. The audience looked at each other, could he be right? How can this be true? Everyone in the audience was invested in solving the challenges of their respective fields with AI. Were they wrong? The professor then clarified this statement ‘you need to have a system that works and that you understand and then you can add AI to improve its effectiveness’. The point he wanted to get across was that it is better to utilize AI to improve a system that already works instead of using it to solve an ill-defined problem that nobody fully understands.
In business and insurance in particular, having a system that works means adopting business models and business processes that you, or someone else, has proven to work. If you know where you are going, then you can be effective in how you get there. In our research at Loughborough University, in collaboration with several partners including Lloyds of London and Willis Tower Watson, we explored the business models used today in insurance that utilized AI and data technologies. After exploring hundreds of insurers from across the globe, 27 case studies were developed. We then found common patterns and narrowed them down to four business models. These four business models capture four different dynamics towards using AI in insurance. While all insurers do not fit perfectly into these four categories, as they can have several services and strategies, the four models capture the different forms of change AI is causing. Neither those that expect it to be business as usual in insurance, nor those that expect a comprehensive disruption, will be fully satisfied.
Figure 1. The four AI and data driven insurance business models
(Adapted from Zarifis A., Holland C.P. & Milne A. (2019))
1) The first model on the left in the diagram is a focus strategy with disaggregation. The insurer confines themselves to a smaller part of the value chain but benefits from the ecosystem around them. This approach is often associated with collaborating with large tech companies or platforms more extensively.
2) The second approach is to avoid changes at business model level and the value chain. Instead, the existing processes are improved with AI and data technologies. This often involves some inhouse research to adapt new AI applications to existing processes.
3) The third model is to expand beyond the current insurance value chain, seek new sources of data and become the platform. Here the insurer extends their business model to utilize the new opportunities offered by AI and big data. The new health tracking services Bupa offers its customers are an example of this.
4) The last approach does not involve an existing insurer but someone coming from outside the sector and disrupting. In many cases this is a tech company expanding into the insurance value chain and offering their own services directly to consumers. An example is Tesla offering insurance to its drivers in some states of the USA (Zarifis, 2020).
While AI can be cognitive and it will change insurance, it is important to choose and apply the right business model to utilize it. An insurer, like an automated vehicle, must have a proven model guiding it with conviction and safety. This model can then be reinforced with extensive use of AI and data technologies.
Zarifis, A. (2020). Why is Tesla selling insurance and what does it mean for drivers?, The Conversation, Available from: https://theconversation.com/why-is-tesla-selling-insurance-and-what-does-it-mean-for-drivers-130910, [Accessed 3rd March 2020].
Zarifis A., Holland C.P. & Milne A. (2019). Evaluating the impact of AI on insurance: The four emerging AI and data driven business models, Emerald Open Research, pp.1-17. Available from: https://emeraldopenresearch.com/articles/1-15. (Open Access)