Insurers are using nascent data signals and critical insights generated in the claims process to fine-tune responses and address anomalies in real-time. (Credit: BritCats Studio/Adobe Stock)
Property and casualty (P&C) insurance business models have been put to the test over the last few years as a combination of sky-high repair costs, longer cycle times and a steady stream of rate increases have driven record levels of customer attrition and strained profitability.
In auto insurance, for example, despite industry-wide premium increases of 11.2%, on average, insurers have been losing five cents for every dollar of premium they charge for much of the past year.
While insurers recognize that simply raising rates is not a sustainable solution to the myriad challenges they face, few have been able to implement the types of structural fixes to their businesses that will enable them to offset these rising costs with improved efficiency.
One area where some leaders are having success, however, is in deploying generative artificial intelligence (GenAI) and advanced analytics to immediately identify and plug holes in their claims cycles.
Real-time claims monitoring
Rather than reinvent their entire operational structures, these insurers are finding ways to tap into nascent data signals and critical insights generated in the claims process to fine-tune responses and address anomalies in real-time.
For example, personal lines auto insurers lose an estimated $30 billion each year due to missing or erroneous underwriting information or other errors that occur in the claims process. By finding ways to spot the anomalies behind those errors, insurers are making it possible to intervene earlier, proactively correct the issue and streamline the entire claims cycle.
The process begins and ends with data. First, insurers must be able to access a broad mix of structured and unstructured data that, historically, has been siloed in different parts of the organization. Modern-day GenAI content extraction tools, however, have made it possible to capture and digest those disparate data sets, which include call logs of customer and agent interactions, claims events and file notes, claim data and proof of loss, policy coverage details and other variables.
Before GenAI, accessing, aggregating and interpreting all of the data would be next to impossible. Today, it can be done in minutes.
Once this data is ingested and organized, it can then be fed into rules-based algorithmic models designed to search for specific triggers and anomalies consistent with problems. These can be things like mismatches between policy coverage details and planned payouts, patterns of poor results with specific repair shops and/or adjusters or even signals of declining customer sentiment based on service interactions.
Continuous performance improvement
At EXL, we recently launched a specialized Insurance Large Language Model (LLM) leveraging NVIDIA AI Enterprise to support these types of claims and underwriting-related tasks. The model was developed to address the challenges many of our insurance customers were having trying to leverage off-the-shelf LLMs to extract and analyze highly specialized insurance data. Because it is trained on specialized data, however, our LLM is consistently achieving a 30% improvement in accuracy on insurance-related tasks over the top pre-trained models, such as GPT4, Claude and Gemini.
In addition to helping us extract meaningful information from across various data silos within the P&C insurance enterprise, the LLM is also used to power a real-time transaction dashboard that keeps claims management teams on top of any red flags or brewing issues that would benefit from intervention. Throughout the process, this system continues tracking each client interaction and each ensuing outcome and feeding that data back into the system to make it smarter.
Most importantly, though, the introduction of GenAI has made it possible to process and analyze this information at scale, allowing insurers to monitor and review 100% of claims processed. Using traditional, manual claims audit processes, most insurers would be lucky to be able to review 5% to 10% of claims.
The result of this comprehensive, real-time approach to claims analytics is a proactive approach to performance improvement. That shows up in the form of more accurate and predictable results, reduced leakage from erroneous payouts and, importantly, a more satisfying customer experience. Research has shown that P&C insurance customer satisfaction improves when customers experience an insurer-initiated rate increase if the customer understands the reasons for that increase in advance. By contrast, when customers are caught off guard by a rate increase, they immediately become a flight risk.
By using GenAI to get a better handle on the data they already have, insurers are able to get in front of the issues that are driving inefficiencies and creating friction with customers. Armed with that intelligence, they can take the necessary steps to improve their own internal workflows and develop more collaborative, constructive relationship with customers.
© Arc, All Rights Reserved. Request academic re-use from www.copyright.com. All other uses, submit a request to TMSalesOperations@arc-network.com. For more information visit Asset & Logo Licensing.