How Insurers can use AI to increase profitable growth

GFT sees the rise in AI affect three keys opportunity areas: new products, customer satisfaction and core underwriting. To understand and leverage this intersection of AI and profitable growth, it’s important to explore these three areas and consider how to build your AI strategy to optimize each of them. 

Changing demographics, consolidation and ever-increasing competition across the insurance space has accelerated over the past five years. This competition places huge time and resource pressures on the traditional innovation process: analyze, design, program, integrate and test.  

Your organization’s ability to create innovative and effective products is a direct indicator of success—consumers gravitate towards products that cut through copy-cat thinking and instead provide integrated value that fits with each customer’s unique lifestyle. Such products earn customer loyalty and lead to longer term relationships that insurers can, in turn, leverage to drive improved profitability.

While growing your customer base through new product adoption is critical to profitable growth, retaining existing ones is just as important. Classical methods of surveys and monitoring of consumer behavior are “table stakes” in today’s cut-throat market. Instead, leading insurers have become expert marketers and data miners across wide swaths of data—Including demographic shifts, social media data and buying patterns.  

Furthermore, leading insurers have retooled their customer experience with the goal of anticipating needs before they even arise. If executed correctly, this virtuous cycle can lead to happier customers and increased profits for insurers.

As a core competency of insurance, it’s not an overstatement to claim that effective underwriting could be the single most important driver of both profitability and success for your firm. Unbalanced claim expenses versus premiums can lead to budget deficits, as can uncontrollable occurrences such as natural disasters and other unforeseen events. Clearly, it’s of paramount performance to create a robust, scalable underwriting process that ensures profit by effectively pricing products.  

While leading insurance providers have leveraged increased automation to reduce manual tasks and free up expert resources and lower overhead, current conditions spell trouble ahead. For example, Mark Englert from ALM Property Casualty 360 states that “Economists predict that, in most markets, the industry will not be able to generate sufficient returns to cover its cost of capital in 2024 or 2025 due to the rising expense of litigation and economic inflation, which leads to higher claims costs.” 1 Englert, M. (2024, February 27). Will the P&C Insurance Market Stabilize in 2024?

As a core competency of insurance, it’s not an overstatement to claim that effective underwriting could be the single most important driver of both profitability and success for your firm. Unbalanced claim expenses versus premiums can lead to budget deficits, as can uncontrollable occurrences such as natural disasters and other unforeseen events. Clearly, it’s of paramount performance to create a robust, scalable underwriting process that ensures profit by effectively pricing products.  

While leading insurance providers have leveraged increased automation to reduce manual tasks and free up expert resources and lower overhead, current conditions spell trouble ahead. For example, Mark Englert from ALM Property Casualty 360 states that “Economists predict that, in most markets, the industry will not be able to generate sufficient returns to cover its cost of capital in 2024 or 2025 due to the rising expense of litigation and economic inflation, which leads to higher claims costs.” 

Artificial Intelligence, including Machine Learning (ML) and Generative AI (GenAI) has swept through every industry and technical area—Insurance is no exception. GFT’s research and experience surfaced a number of ways that insurers should consider using the technology specifically to augment profitability improvement efforts, including:  

Effective AI cannot be achieved without effective data management. As insurance firms grapple with data consolidation, quality and lineage, AI tools can accelerate how data is integrated, cleansed, and harmonized.  

In addition, new technologies such as Vector Databases can represent multiple data types (i.e., sensor data, words, images, and audio) and allow them to be searched simultaneously. Large Language Models (LLMs) can be leveraged to understand and reason over huge amounts of data, helping insurers to both find patterns in existing data and refine insurance products to specific market segments. Two possible examples of this in action could be: 

• Parametric Insurance: This includes policies that pay out a predetermined amount of money upon the occurrence of a specific event, as defined by pre-agreed-upon parameters, such as seismic activity, weather conditions, or financial indices reaching predefined thresholds. 

• Usage-based Insurance: This is an auto insurance product where premiums are based on individual driving behavior and usage patterns rather than traditional factors such as age, gender or location. 

AI brings new possibilities to depth and breadth to the challenges of knowing customers, anticipating their needs and optimizing the Customer Experience (CX). Coupled with next-generation Data Lakes and Data Mesh-approaches and near-infinite storage, the long sought-after 360-degree view of the customer is finally achievable. 

We have all seen -fueled customer sentiment analyses in chatbots and other interactive channels. The logical next steps are to further enhance CX to combine internal customer feedback with the rest of the front and back-office data and even external data services and products. The resulting federated data sets can then be used to train smaller LLMs for specific purposes. The end result will provide insurers with highly personalized views of their customer bases, leading to increased Customer Satisfaction (CSAT) scores and increased long-term repeat business. This investment leads to Insurers having increased options for specialization, monetization, and profit possibilities.  

Modern underwriters already use Optical Character Recognition (OCR) and Natural Language Processing (NLP) as routine tools to extract essential information from unstructured data sources and documents. As above, GFT believes that the tsunami of growth in data volumes and the maturation of GenAI and associated technologies will only accelerate the benefits for leading underwriting firms. 

Obvious expansions of this include historical claims data and public records. Additionally, specialized LLMs that understand medical history and lifestyle choices will become commonplace and help insurers to accurately predict health risks. Assuming data privacy, opt-in mechanisms and other critical security standards continue to protect consumers, insurers will be able to tailor policies and costs in favorable (but fair) ways. 

With continuous advancements with AI and so many possibilities and options, it is sometimes challenging to know how to get started. 

Based on hundreds of AI projects across insurance, financial services and manufacturing, we suggest these initial guardrails: 

A robust data management strategy is essential. Insurers must ensure data quality, accessibility and security to extract insights from data sets. This need not be perfect from day one but is a prerequisite if you expect success in specific subject areas. As a recent Gartner article suggests: 

With continuous advancements with AI and so many possibilities and options, it is sometimes challenging to know how to get started. 

Based on hundreds of AI projects across insurance, financial services and manufacturing, we suggest these initial guardrails: 

A robust data management strategy is essential. Insurers must ensure data quality, accessibility and security to extract insights from data sets. This need not be perfect from day one but is a prerequisite if you expect success in specific subject areas. As a recent Gartner article suggests: