Six Emerging Ways for Insurers to Realize Value from AI

Cloud technology, new data sources and advances in algorithmic methods are enabling insurers to make speedy progress in simplifying customer experience, improving underwriting and optimizing portfolio performance.

(Image credit: Shutterstock.)

Artificial intelligence (AI) offers great promise, especially for the insurance industry. Proliferation of cloud technology, availability of a number of emerging data sources and advances in algorithmic methods are enabling insurers to make speedy progress in simplifying the customer and claims experience, including understanding new risk pools, improving underwriting and optimizing portfolio performance. As much as AI experts talk about the technology’s benefits, it’s not always clear how insurance companies can put AI to work to in order to deliver outcomes. With that in mind, here are six emerging ideas for realizing value out of AI for insurance companies.

#1: Transform underwriting

The insurance underwriting process has arguably the greatest potential for transformation by AI. Text mining offers one of the most interesting solutions to improving underwriting processes and outcomes. Our recent work indicates that there is potential to reduce the time to draft policies by 20 to 25 percent and even higher when reviewing changes in wordings within policies, e.g. specific clauses, specific exclusions, etc. Tremendous value can be gained by also looking at submissions that didn’t end up getting bound, similar to what banks do on ‘reject inferencing.’ And further, by combining external data with internal claims data, insurers can improve pricing for new business. For example, a life insurance company can incorporate data from gym club memberships, travel plans, driving habits, and hobbies to make more informed underwriting decisions based on predicted risk and the customer’s needs. In the future, large portions of the underwriting process—which are manual today—may be fully automated with AI.

#2: Improve the customer experience and deliver next-best actions

In highly competitive industries like insurance, customer service is a key differentiator. AI can give insurance companies the advantage they need to improve the customer experience. For example, AI can be used to identify the factors that contribute to a positive customer interaction and proactively engage at-risk customers to reduce churn. It can also personalize product recommendations and next-best actions based on the customer’s profile. Arguably one of the highest opportunity areas to apply AI within Insurance, across most lines of businesses, is in driving customer engagement and experience. From better cross-sell, personalized targeting in personal lines to climate insights for farmers to geospatial insights for marine and other specialty insurance clients, there are significant engagement and experience outcomes that can be driven through advanced analytics.

#3: Reduce customer friction and the cost to serve

One of the best ways to reduce customer friction is to reduce or eliminate wait times. That’s one reason why both companies and their customers prefer self-service channels over live customer-service agents, especially the younger and digitally savvy population. Self-service can be faster and less expensive—if the self-service channel is as effective as a live agent. With the help of AI, it can be. AI can help improve the effectiveness of chatbots and virtual agents by enabling them to learn as they respond to requests.  Our recent experience suggests that there is considerable opportunity to reduce friction that customers face especially on the digital assets (e.g. websites) of insurance companies through a three step approach of sensorizing the various parts of the digital assets that cause breaks in customer journey, applying sophisticated algorithms for anomaly detection and root cause identification and A/B testing at scale, insurers can aim to reduce the cost of servicing customers, especially that of contact centers and optimize processes.

#4: Go beyond proofs of concept and apply AI across the insurance value chain

AI may still seem futuristic, but the fact is the technology is here, and it’s ready for prime time. With increasing competition, and advances in the above enablers, insurers need to deploy a fail fast approach and run multiple rapid Minimum Viable Proposition (‘MVP’) initiatives, in parallel. The most impactful emerging sources of internal data from our recent experience include Voice of Customer data (calls, social, etc.) and telematics., Our recent experience suggests that about 25-30% of MVPs will result in eventual operationalization at enterprise level, which makes it critical to have a fail-fast mentality. The insurance industry has historically collected a wealth of data, making it an ideal candidate for using AI across the value chain, whether on traditional use cases such as pricing, claims analytics, customer segmentation, marketing & sales effectiveness or on emerging areas such as hyper-personalization, understanding cyber risk, automated underwriting, etc. whether through active customer segmentation for marketing purposes or by directing customers through their preferred sales channel.

#5: Beyond AI: Operationalize through better engineering and behavioral sciences

AI alone is not enough. This is perhaps the most fundamental insight we gained from working with the largest and most complex insurers across the insurance value chain. Insurers are plagued with a big challenge which is to overhaul the technology environment that they are having to work with. A 2019 Gartner CIO survey indicates that most P&C and life insurers are still building out the foundation for analytics and facing legacy system challenges. A few leading insurers have recognized the magnitude of the challenge and started building data engineering teams or working with ‘acceleration’ partners to build data lakes, run cloud migration programs and setup better downstream processes to simplify consumption of AI outputs.

An important consideration for the insurance industry is handling regulations. And with more sophisticated approaches emerging, the challenge of improving transparency and accountability of AI methods is becoming important to manage. Add to it, the fears of AI adoption and cultural re-orientation that the organization needs to go through to understand and accept that the opportunity through AI is real, and that AI will help them be more effective in their roles. This requires a behavioral sciences based approach to empathize, up-skill, and engage the people that need to act on the insights driven by AI, i.e. the underwriters, policy administrators, claims managers, contact center agents, etc.

#6: Make measurement a priority 

Digitizing business processes and infusing AI to improve them enables the impact to be highly measurable. Historical data can be captured from many sources to form a baseline for gauging progress. These practices enable insurance companies to quantify cost reductions, productivity improvements, and other outcomes they realize from AI to help make a business case for its use. In our experience, the philosophy towards measurement is critical to get right. It is important to internalize that the value of AI is not realized from the methods or number of models or statistical uplifts, but from the execution of actions coming out of these models. And oftentimes, the best pilots to scale may not be the ones with the greatest estimated benefit, but the ones that are simplest to execute (i.e., requires minimum change to operating process). It should also be cautioned not to attribute outcomes solely to AI, as most opportunities and problems in the Insurance world are connected and as the saying goes, “it takes a village.

In summary, the insurance industry is poised to benefit from AI, and the technology is available to do so. The hope is that the six points above can be used to enable a dialogue within an organization, so that the company can envision how to realize AI’s benefits at an enterprise level.

AI and New Technology Adoption in Commercial Lines and Workers Comp



Sankar Narayanan // Sankar Narayanan is a member of the executive leadership team at Fractal Analytics and Business Head for BFSI, TMT, Healthcare & Engineering verticals, with P&L responsibility for Consulting, Sales & Delivery teams. Narayanan has over 15 years of multi-market experience in helping more than 100 organizations achieve measurable top line/bottom line impact through analytics. He has held positions with IBM, SunGard and Sallie Mae Corp. in business analysis & product development. Narayanan has an MBA from Indian Institute of Management, Ahmedabad and Bachelor of Engineering, Control Systems from Madras University, Big Data Opportunities, Challenges & Applications – course from Massachusetts Institute of Technology.

Leave a Comment