 By Alice Underwood, Global leader, Insurance Consulting and Technology and Tom Helm, Head of Claims Consulting from Willis Towers Watson
And yet, the claims function has thus far remained relatively untouched by the advances in automation and analytics that have transformed actuarial processes. That is starting to change and leading-edge insurers are now building out their claims analytics capabilities. By embracing data science, actuarial insights, automation and artificial intelligence, the industry’s early adopters are already seeing improved cost control, increased fraud detection, reduced claim life cycles and better customer outcomes.
What’s possible?
A number of players in the new ClaimsTech market have made exciting breakthroughs. Examples include Snapsheet, a smartphone app enabling US motor insurance customers to handle their claims via smartphone, potentially going from “photo submission to cost estimate in 2.7 hours, with claims closed in 2.5 days and a customer satisfaction rating of 9/10.” RightIndem is a customer-oriented claims platform that can be customised by insurers and MGAs, which they say enables their clients to “provide a market-leading claims experience, as the service links all partners in the claims system and is supported by a series of AIs that can reduce indemnity service cost by 10% while shortening cycle time”.
Established insurers can take advantage of these and other start-up offerings and/or develop their in-house capabilities to improve claims processing. Leading multinational insurer Zurich is embracing AI technology in claims and is already using AI to extract relevant information from claimant medical reports. Zurich states that this is delivering improved accuracy and has also “saved 40,000 work hours, while speeding up the claims process time to five seconds”.
The Internet of Things can also help improve claims service. Insure the Box, a provider of telematics-based motor insurance in the UK, has used the technology and data supplied by their partner Octo Telematics to develop an accident alert system. If the “box” in their customer’s vehicle records a certain level of G-force, claims handlers are alerted to call the customer — often at the scene of the accident — to help with the vehicle recovery and claim process.
Predictive modelling tools can help drive proactivity across the claims process. Traditionally these tools have been used in pricing, reserving and fraud detection; but they can also be utilised to provide a wide range of decision support and intervention solutions within the claims functions including, for example, predicting litigation probabilities and identifying claims for which the customer experience is at risk of falling below the desired standard.
Recently the team at Willis Towers Watson developed such predictive models for workers’ compensation carriers, using structured and unstructured data to identify problem claims likely to unexpectedly “jump” in their incurred cost. Attempting to identify these claims manually is like trying to find a needle in a haystack, but with this predictive modelling approach, insurers can take action much earlier — providing a better claimant experience and managing indemnity costs.
Data fuels the engine
A major advantage that established insurers enjoy over InsurTech start-ups is their historical data asset. However, in many cases this data has been captured in legacy systems having variable degrees of consistency and quality. This “weak” data is often the first target when taking advantage of advanced modelling techniques. For example, unstructured text or voice data can be used to augment historic structured data; given the level of insight that can be gained from even a limited amount of voice or text information, this kind of data augmentation can bring exceptional value to insurers, helping to shine a new light on their claims history and trends.
Real-time decision engine
The “need for speed” will be a cry from claims teams working with integrated automated solutions, particularly those operating in classes such as motor or household in which service levels and cost control are time critical. To support or fully automate decision making requires a sophisticated and powerful engine which can run complex models and deliver the outputs of analytics to handlers in real time.
Increasingly, insurers find that actuarial software originally developed for pricing decision support can also provide decision support for claims handling as well. This sophisticated software is a great fit as it was designed to integrate with other systems (such as claims handling systems) and process complex models (such as GLM and machine learning risk cost models) within sub-second response times to deliver real-time pricing and underwriting outputs. Crucially for claims use, the software is scalable, having already been built to cope with millions of aggregator quote requests per day.
A case in point
One insurer in Europe is pursuing this path with vigour. Working with claims experts at Willis Towers Watson, analysts and data scientists, their focus is on developing machine learning models to improve fraudulent claims detection. These models are then deployed into the claims operation via Radar Live, our real-time decision engine. With an increasing proportion of the insurer’s customers reporting their claims online, this approach allows the models to run and flag in real-time any cases in need of specialist investigation. Because this insurer was already using Radar Live as their agile real-time engine for pricing decision support, they could simply broaden the application of this familiar platform — which not only reduces time to deployment but also provides meaningful cost savings. And, given their significant advanced analytics capability in house, this insurer can control and update the predictive models themselves as required.
Mastering the balancing act
Automating claims decisions is not easy. There are often complex considerations that require an element of human judgment. When helping insurers through implementation, it is important to remain vigilant to unintended consequences. For example, improved efficiency and accelerated claims payments must be considered when setting reserves and updating rates; and such improvements should not come at the expense of weakening fraud control.
Executed well, this wave of AI and automation sweeping through the claims space offers insurers the opportunity to transform many material aspects of their operations, delivering a service that delights their customers, whilst also reducing claims costs and loss ratios.
|
