The use of artificial intelligence (AI) in insurance underwriting is a proven and powerful tool for enhancing growth and profitability. AI enables insurers to detect misrepresentation and policy fraud quickly and accurately before a policy is bound. Here is a summary of two case studies demonstrating the application of AI in underwriting and the results achieved.
The power of AI in insurance underwriting
By helping insurers more accurately assess risk before policies are issued, the use of AI in underwriting has helped leading insurers set new benchmarks for risk mitigation. By detecting risk in this critical stage, these technologies can ultimately reduce exposure to high-risk policies and minimize future claims losses.
Enhancing underwriting approaches a can have a profound impact on the combined ratio. In fact, Shift's experience working with leading insurers has shown that mitigating fraud during the underwriting process can result in shaving up to five points from the combined ratio.
Proven benefits of AI in insurance underwriting:
- Enable faster decisions
- Grow the book of business confidently
- Increase in accuracy of applications flagged for review
- Improve rating accuracy
- Avoid unnecessary rate increases
- Improve customer satisfaction and retention
Case Study 1: Top 5 U.S. insurer adopts artificial intelligence in insurance underwriting for projected $30M+ in mitigation annually
Situation
In the competitive world of personal auto insurance, customer experience is critical to success. A top 5 P&C insurer in the U.S. knew they couldn’t let hidden fraud and risk put customer satisfaction in jeopardy, so their fraud task force launched an initiative to identify innovative fraud prevention solutions for Underwriting. As a Shift Claims Fraud customer since 2021, the insurer was well aware of the power of advanced AI in detecting claims fraud. However, the Insurer’s fraud task force and Underwriting teams were focused on:
- More efficient risk detection: The insurer was looking to stop more fraud and hidden risk on new policies while maintaining current underwriting staff levels
- Fraud networks: The insurer recognized that misrepresented auto policies sold on to unsuspecting consumers by “ghost brokers” could jeopardize customer satisfaction and their brand reputation, while resulting in claims with an average 500% loss ratio
The Underwriting leaders also knew that they needed a solution that could stop fraud and hidden risk before costly claims were paid. That meant daily analysis during the new business “free look” period was critical, while still maintaining the insurer’s underwriting process for fraud and misrepresentation review to ensure customer satisfaction. That way, the insurer could adjust the risk tier, cancel fraudulent policies, or intensify account monitoring.
The solution
With the combined expertise of the Insurer and Shift, a proof of concept was initiated on 2.8 million auto policies. Shift’s data scientists optimized and deployed AI detection models for underwriting risk, while Shift’s Underwriting subject matter experts proposed investigation guidelines and workflows. Shift’s implementation included:
- Misrepresentation detection: Risk detection algorithms specifically designed to analyze new policies in order to catch fraud and misrepresentation during the “free look” period
- Ghost broking and network detection: Shift implemented powerful network analysis AI to uncover patterns of seemingly “normal” policies connected to ghost brokers
- Workflow optimization: Shift’s Underwriting experts worked with the insurer to integrate Underwriting risk detection into their existing workflows, with 100% explainability and complete audit trail, so that the insurer could stop fraud with their existing Underwriting team
- Unified fraud detection across policy and claims: With both Underwriting Risk and Claims Fraud implemented, Shift’s entity resolution AI uncovered policyholders using fraudulent information to hide prior claims history
The result
Shift Underwriting Risk generated more than $15 for every new policy in incremental prevented losses, projected at more than $30M USD in underwriting mitigation annually, all achievable while maintaining Underwriting staff levels. This proof of value helped the insurer realize the potential for Shift’s AI to support their customer growth and satisfaction goals.
- 40% impact rate on policy alerts
- $30M+ in annual projected underwriting risk mitigation
- 500% average fraud network loss ratios avoided
Download a pdf version of this case study
Case Study 2: European P&C insurer implements underwriting AI for 3x the expected relevant policy risk alerts
Situation
The head of innovation at one of Shift’s European P&C clients recently began a review across their business to identify applications for Shift’s AI decisioning capabilities. Having seen the impact of Shift Claims Fraud Detection, the insurer’s head of policy service asked about the potential to address challenges for their underwriting team:
- Missed premium risk: The firm’s legacy underwriting process renewed risky or potentially fraudulent policies without any safeguard for fraud or risk
- Inefficient manual investigation: End users spent too much time operating manual controls and performing manual underwriting reviews
The policy services executive needed a solution that could scale the insurer’s underwriting practices in order to better identify fraudulent or misrepresented policies in their portfolio. At the same time, they still needed to accelerate the insurer’s underwriting process to maintain customer satisfaction. Ideally, the insurer would analyze policies in order to adjust risk tier, cancel fraudulent policies, or intensify account monitoring.
The solution
Based on Shift’s capabilities, the innovation and policy services leadership began a five month test to prove Shift’s value. As a net new capability, the insurer set the criteria of 1 relevant policy alert for every 400 policies analyzed, a minimum of 50% accuracy in risk detection, and automated investigation into underwriting workflows. They also asked Shift to identify novel forms of risk. Shift’s team took on the challenge. In just 3 months, Shift data scientists optimized AI detection models for underwriting risk, enhancing the insurer’s guidelines and workflows. Shift’s implementation included three key features:
- AI decisioning built for underwriting: Risk detection algorithms were specifically designed to analyze policies after they were bound in order to catch fraud and misrepresentation
- Expanded detection: Shift configured scenarios from its AI decisioning library to cover a broader scope of suspicious patterns and unusual situations
- Workflow integration: Shift optimized risk detection to automatically generate alerts for the firm’s unique policies and catch high-priority fraud scenarios, with 100% explainable context to support underwriting risk decisions
The result
Over the 5-month proof of concept, Shift Underwriting Risk generated more than 3x the expected relevant policy risk alerts, while also exceeding the 50% accuracy benchmark. The results led to at least $1M USD in underwriting impact when applied across the portfolio, through reduced premium leakage, account cancellation due to fraud risk, or policy surveillance.
As an extra proof point, Shift’s AI uncovered hundreds of accurately identified new cases of policy risk where the underwriting team had already manually reviewed and missed, highlighting the power of AI to the innovation and policy services leadership.
This proof of value helped the insurer realize the potential for Shift to make a difference during the underwriting process. Satisfied with performance, the head of policy services expanded Shifts Underwriting Risk Detection across their personal auto line of business.
Now fully implemented for auto policies, Shift Underwriting Risk is accelerating the underwriting process for the insurer, providing scheduled policy alerts at the time of new business, claims, and pre-renewal, with complete alert context to streamline investigations.
Download a pdf version of this case study
- $1M in annual projected underwriting risk mitigation
- 3x more risky policies investigated
- 100+ Shift identified risky policies
- Exceeded 50% accuracy benchmark
