• Home  >  
  • Perspectives  >  
  • How Data Science is Reimagining the Future of Insurance  
Blog July 7, 2021
4 min read

How Data Science is Reimagining the Future of Insurance

Author: Minu Philip Until recently, Insurance has largely been perceived to be a traditional and slow-moving industry. However, the entire […]


Author: Minu Philip

Until recently, Insurance has largely been perceived to be a traditional and slow-moving industry. However, the entire sector is now experiencing significant disruptions led by insurtech, climate-change-induced natural disasters, and, of course, COVID-19. Each of these forces has added unique challenges to the insurance industry that is still undergoing its digital renaissance.

The Reshaping Forces

Insurtech companies are leading the way in redefining the Insurance value chain right from targeted product concepts to end-to-end automation. For instance, there are new products like pet Insurance and customized coverage (hourly car Insurance by Cuvva). On-demand insurers like Trov offer consumers a mobile-enabled on-off switch for coverage. ClaimDi has eliminated the need for manual claim adjusters with their mobile app. Instead, claimants now interact with insurers on the accident site just by shaking their phones.

While traditional Insurance companies were just beginning to see value in this digital-led transformation, climate change accelerated the need for the transition and COVID-19 added an urgency. The increasing frequency and intensity of natural calamities are forcing Insurance companies to not just look at risk measurement and transfer, but also risk mitigation. This will involve working with customers and governments to effectively manage and maybe even prevent such risks.

The COVID-19 pandemic has also highlighted the importance of being prepared for catastrophic events – not just in terms of minimizing losses but also in building business processes that can function and scale without relying on offline channels. Insurance companies and agents accustomed to in-person interactions are rapidly evolving to provide uninterrupted online service to customers. Insurance agents are also rethinking how they can build client relationships through online channels given global lockdowns. A survey of US agents by McKinsey in January 2020 showed that about 90% of Life Insurance agents’ sales conversations and nearly 70% of their ongoing client conversations were conducted in person. In the follow-up survey in May, less than 5% of agents had any in-person conversations. A late-April 2020 survey of European Insurance executives found that some 89% of respondents expect a significant acceleration in digitization, and most also anticipate a further shift in channel mix.

State of Insurance in the Next Decade

Addressing the above challenges involves transforming the entire Insurance value chain. Some key changes that are happening are:

Product & Distribution

– Innovative & custom coverage options
– Online distribution channels that are becoming completely self-serve

Pricing & Underwriting

– Continuous underwriting model that is replacing one-time pricing at the point of first purchase
– Straight through processing for applications, including fluidless (no lab tests needed) processing for Life Insurance

Loss & Claims Management

– Shifting focus to loss prevention rather than loss adjudication with IoT playing a key role
– AI-driven loss measurement and claims management processes

The Role of Data Science

In all of this, what is our role as Data Scientists? Insurance companies are looking to us to answer two key questions.

1. What is the right data that can help them in this transformation? This includes identifying different types of external data that can add value, sourcing this data, establishing ground truth, and assessing the ROI

2. How can they effectively use this data in decision-making? Here, the reference is to not just use the ML algorithms but also to see how we can create end-to-end decision frameworks and platforms, ensuring that we are doing the right thing by the customers and the business while abiding by regulations.

A good example of a transformative data science solution is how external data is being leveraged to improve the commercial underwriting process. Until now, this has been a cumbersome process for certain types of coverage, which involved applicants filling in a long questionnaire of 40-50 questions which then had to be validated by underwriters. As could be expected, the process used to see a lot of drop-offs and incorrect data. The question was how to simplify this process for the applicant at the same time ensuring the underwriter had enough information to make the right decision.

This is where we start looking beyond the traditional enterprise data to see how we can utilize external data sources to enrich our understanding of business applicants and prefill their applications. There is a lot of information available in the public space on businesses – ranging from data providers like Better Business Bureau that collect certain information in a structured way to primarily consumer-driven information on platforms like Yelp. Building a robust solution based on these sources involved a lot of exploration and experimentation to ensure that we are bringing in accurate and meaningful information to drive a critical business decision such as underwriting. The key questions addressed by this solution include:

– How to merge information across sources in the absence of a common identifier?
– How to extract meaningful information from customer reviews and images that could indicate risk levels for the business?
– How to establish ground truth and assess the value of different data sources to help in investment decisions and prioritization?

Additional information on the use case can be found here.

Being a responsible change partner

While it is an exciting and challenging time to be a Data Scientist in the Insurance domain, it brings with it a lot of responsibilities as well. Given that Insurance is a heavily regulated industry, it is important to ensure that all data-driven decision frameworks are explainable and use only those data elements that will pass the regulatory scrutiny. Added to this is the responsibility to ensure that the models are not perpetuating any hidden biases or discriminations. This requires setting up strong Model Risk Management frameworks that can define, assess and monitor the different risks associated with the model.

Overall, it is transformation time for Insurance. As companies evolve to adapt to the technology and environmental changes, it will no doubt herald a new era in the industry.

Stay tuned to delve deeper into the reshaping forces and the future of Insurance, the role of data science, and the necessity for Model Risk Management frameworks.

Copyright © 2024 Tiger Analytics | All Rights Reserved