New AI imperatives in claims for 2022
Here are three factors driving AI transformation in insurance claims.
In 2021, the insurance industry was buffeted by the four-pronged effects of climate change, social change, the pandemic, and the evolving expectations of employees and policyholders. In the year ahead, we expect these forces to continue to accelerate digital transformation.
First, profit pressure (particularly from cat losses) will keep insurers focused on finding new ways to reduce loss ratios and price risk accurately. According to investment bank Jefferies, cat losses in 2021 were the fourth highest annual total in the last two decades and the highest in the last three years.
Second, we saw a relationship between the pandemic and potential fraud. For example, looking at auto insurance, while most insurers saw a decrease in physical/collision auto claims due to lock-down scenarios, the increase in comprehensive claims from stolen, abandoned or mysteriously burned vehicles rose. Because of this, adjusters had to re-evaluate the “pressure” aspect of many claims, which can be one of the most challenging elements of an investigation. For many carriers, that, combined with a growing consumer demand for immediate results, left them in a presser-cooker type of scenario in need of immediate action.
Third, financial services and insurance were among the top four industries that saw the highest increases in teleworking during the pandemic. In the year ahead, we expect that insurers will increasingly see remote work and its related digital initiatives as a more permanent way to attract and retain talent, not just a grudgingly accepted short-term response to lock-downs.
The trends in remote work are compelling. According to Mercer’s 2021 Flexible Working Policies & Practices Survey, 70% of companies said they were planning to adopt a hybrid model allowing flexibility to both work at home and in the office. In addition, Microsoft’s 2021 Work Trend Index found that 66% of these employers are already redesigning their workplaces to accommodate hybrid work arrangements. As insurers compete against other employers including big tech for talent in 2022, the case for remote work will reach higher levels of urgency. The FlexJobs 10th Annual Survey confirms that 58% of the labor force report wanting to be full-time remote employees post-pandemic, while 39% want a hybrid work environment. In short, 97% of workers expect some form of remote work.
For claims operations, the three forces of change pose very unique challenges. Throughout the claims journey, specialized knowledge of regional regulatory rules ― not to mention experience and heuristics acquired over several years ― must be reinforced in claims professionals. Claims leaders are thinking about opportunities to automate workflows, empower claims professionals with decision support tools and data, and connect policyholder experiences to the back office.
Imperative #1: Digitally redesign end-to-end claims journeys
Out of the six phases of the claims journey, insurers are starting to ramp up in two traditionally under-invested areas: claims recovery and special investigations units (SIU). It would be helpful to view these two areas in context.
Imperative #2: Improve claims recovery to reduce loss ratios
Claims recovery encompasses subrogation and salvage, processes that enable insurers to offset paid losses by pursuing hard dollars from liable third parties. Even though claims recovery could materially reduce loss ratios, insurers have traditionally regarded this area as a “backwater” of claims operations. Until recently, very little was done to automate and apply machine-learning to subrogation. Insurers and their subro vendors have by and large identified recovery potential and pursued files with manual human power. As a result, insurers experience leakage in their claims processes and miss opportunities to increase net recoveries. By some estimates, the gap between today’s recovery rates and improved recovery rates could amount to tens of billions of dollars per year in the U.S. property and casualty market alone.
A national commercial insurer decided to solve these problems in subrogation by applying machine learning to detect and quantify recovery potential and digitize the workflows to pursue those dollars. Prior to its recovery modernization effort, it expended significant human power to manually review claims. Recovery professionals had to manually review as many as over 20 files to arrive at one economically viable file. After the solution, the insurer was able to cut false positives by over half, increase its net recovery potential by over 15%, and profitably pursue recoveries on three-quarters of flagged files.
Imperative #3: Double-down on fraud prevention
Insurers are beginning to realize that implementing fraud solutions would allow them to complete more quality investigations faster, while simultaneously maintaining the best possible customer experience for the honest 90% of their insureds. These safety nets encompass the entire claims lifecycle with targeted screenings at pivotal moments within the claim. They identify both claims that are eligible for straight-through-processing (STP), and those that need further triage by the claims adjusters.
Prior to launching its fraud solution, an auto insurer built in-house solutions for identifying claims fraud with their team of expert data scientists. However, due to scalability issues, only 10% of their claims volume ran through their in-house fraud models. Further, the insurer wanted to make sure that every claim received was reviewed for fraud with 100% accuracy, consistency and in real-time.
With its AI-powered fraud solution today, this insurer is now receiving results in under two seconds. On top of that, they recognized a 75% reduction in false positives (claims identified as fraudulent by the models but proven to be honest by the investigators) and an efficiency gain of 50%. In essence, they went from a claims fraud screening rate of 10% to 100%. This reduced their average claims resolution time by 50% while increasing customer NPS scores.
How to avoid pitfalls in 2022
There are many positives to AI-driven solutions in fraud and subrogation. Many insurers in both personal lines and commercial lines are achieving STP rates north of 70%. In SIU, insurers are recognizing a total increase in their proven fraud rate of 1-2% (industry standard for proven fraud rate is 3-5%, as stated by the Coalition Against Insurance Fraud). And in claims recovery, best-in-class subrogation solutions are increasing net recoveries by as much as 20% per year, resulting in a reduction in loss ratios by as much as 2%.
However, there are concerns about AI that insurers and insurtechs alike should proactively address. First, employees and policyholders should be assured that AI technology will not replace humans but empower them through decision support and streamlined workflow.
Second, insurers must deal head-on with potential human and legacy process biases that may be inherent in datasets used to train predictive models. Policyholders will want to know that models that affect the disposition of their claims (or the rating of their policies) will be fair, consistent and honest.
Third, insurers should select solution partners who possess rich domain expertise and a track record, not just technological prowess. For example, some AI firms may tack on a subrogation capability as an afterthought with little regard to workflow nuances, regulatory know-how or specialized expertise that is required to deliver tangible business outcomes across business lines.
Finally, insurers should bear in mind the constraints of compliance on innovation. Insurers and insurtechs must continue to be held accountable to standard practices for data protection, data privacy, data usage, and data transfer.
Jeff To is senior vice president of Safekeep-CCC Intelligent Solutions. Jeroen Morrenhof is co-founder and CEO of Friss.
Related: