The UK insurance market is currently suffering from one of the highest claims costs levels in the whole of Europe, coming in at a staggering £9.2 billion in 2016. This translates to spending 63% higher towards total claims costs compared to its European counterpart, and up to 80% of their average total premiums.[i] The main reasons for this are to be found in the variety of participants within the structure of the UK claims market as well as the high level of fraud (read more). However, with the Financial Conduct Authority currently looking closely into the UK Private Motor Claims Market, these inefficiencies are likely to change in the near future. In order to prepare for the challenges ahead, insurers need to internally optimise the effectiveness of claims organisations by focussing on indemnity leakages all the more.
Most notable causes of leakages include insufficient claims assessment accuracy and detection of fraud or exaggerated claims – with fraud alone costing the industry an estimated £3.4 billion per year, around £50 per premium.[ii] As we have seen incumbents suffer from claims leakage of well above 20%, it is irrefutable that accurate assessments of claims severity and validity are a crucial lever for optimising the economic performance for many insurers.
In addressing the challenges to control indemnity spend, new technological levers are becoming increasingly vital. Indeed, our accumulated experience shows that the technologies introduced below can help insurers to increase the accuracy of indemnity spend by 20-30% through better fraud detection, better control of costs incurred at fulfilment stage, as well as a reduction in legal actions.
Whilst we have briefly introduced several levers that can enhance claims effectiveness in our overview article, we will now take a closer look at their potential to increase the control of claims costs along the value chain.
OPPORTUNITIES TO ENHANCE EFFECTIVENESS ALONG THE CLAIMS VALUE CHAIN
FRAUD DETECTION AT THE FIRST NOTICE OF LOSS (FNOL) – Enhance manual processes through data driven automation
Today, FNOLs are mostly received over the phone and the respective evidence is then submitted in various unstructured formats, such as letters, pictures and emails. Unfortunately, more often than not, insurers still store the information in an unstructured way, making it difficult to search through and link the collected claims data. As a consequence, claims adjusters will have to form decisions subjectively based on their training and experience, using past claims in their practice groups as reference. The reliance on manual claims intake and assessment, however, renders human error, subjectivity, and, ultimately, claims leakage unavoidable. To tackle that problem, claims information needs to be translated and stored in a structured way – for example, by means of deploying digital FNOL tools, translating letters with the help of optical character recognition (OCR), or recording, transcribing, and storing phone calls. Once claims data is prepared for digital processing, an array of opportunities to increase effectiveness will open up:
Automated fraud detection: By applying advanced analytics early on in the value chain, for example, insurers have a better chance of flagging up mendacious activities, overall mitigating the risks of further processing a fraudulent claim. Machine-Learning (ML) algorithms are becoming increasingly better at detecting otherwise obscure patterns that human insurance agents might not, especially given the increasing quantity of data available beyond the basic claims information, with external data sources opening up possibilities to garner more and better insights into consumer behaviour and profile. Here, applications to cross-reference different data bases in order to filter out claimants that are prone to fraudulent behaviour are proving to be especially helpful. Other applications of deep learning algorithms in fraud detection range from lexical analysis for ‘hit words’ and photo alteration detection, to facial recognition and voice analysis.
The automation of fraud detection is especially important in order to secure straight-through processing rates without losing control of indemnity spend. What’s more, the application of technologies in recording and assessing claims allows for enhanced effectiveness without increasing the complexity for claimants, mitigating the traditional trade-off between claims effectiveness and customer-satisfaction. Overall, we estimate a potential to reduce claims leakage by 5-10% with the help of enhanced fraud detection.
Examples of suppliers include: Dence digital evidence for cross-referencing data; Verisk Analytics for picture alteration; nemesysco for voice analysis
INCREASED ACCURACY DURING LOSS ASSESSMENT – Support manual processes with smart assessment tools
Similarly to the automated detection of fraud at FNOL stage, the deployment of technology in claims estimation can further reduce leakage from human error and poor decision-making. From our assessment, successful implementations of the levers discussed below can reduce leakage by a further 5%.
Automated claims estimation: AI-solutions can analyse claims information in all formats, including pictures and videos, to generate an initial estimation of claims costs. The main advantage is that those systems can spot patterns that were previously undetected by traditional means of calculations, and therefore allow for enhanced decision making. The accuracy of those predictions, however, depends on the level of training and the amount of data fed into the AI, so that at the beginning, the deployment of AI-driven estimations will be restricted to serve as a support for manual processes. For example, they could be leveraged in order to provide an objective reference as a basis for claims handlers’ manual assessments. However, we believe that AI systems will be able to outperform human adjusters in the near future, rendering these systems a key competitive factor.
Examples of suppliers include: Verisk Analytics; Tractable
Enhanced data insights for reserves: An accurate setting of claims reserves is paramount both internally – to quantify the insurer’s liabilities in order to properly appraise its financial position – and externally to meet regulatory requirements. As it currently stands, human agents are inundated by the variety of complex and often unstructured data, rendering reserves not as precise as they could be. Enhanced estimation tools can improve an incumbent’s control over costs incurred in real-time, especially with regards to better modelling of payment spans for more complex claims.
Examples of suppliers include: Miliman Arius; Towers Watson ResQ
FULFILMENT/REPAIR – Mitigate risks of fraudulent invoices and minimise fulfilment costs
The fulfilment stage of the value chain is characterised by interactions with a vast range of suppliers and entities, e.g. third-party administrators, repair shops, engineers, materials suppliers, pharmaceutical reimbursement managers, and defence counsels. Especially in the UK, these interactions inhibit the potential to stack up avoidable expenses due to costly referral systems between service providing entities. zeb research shows that UK insurers currently pay on average up to 50% more for vehicle repairs compared to their counterparts in France. To prevent repairers from synthetically fixing prices to distribute contracts amongst themselves, incumbents have previously introduced a repair service auction system. However, this has occasionally been met with industrial embargoes, as providers felt insurers were directly agitating the competitive balance within their trade. With digital, real-time tracking of transactions, insurers can still benefit from the transparent cost control and integrated price comparison social Darwinism promises, without explicitly provoking the competitive equilibrium between repair providers.
Digital supplier management: Traditionally, insurers have been dependent on loyal supplier networks, providing a regular flow of business to service providers in exchange for consistency in quality as well as discounted rates. The digitisation of these relationships through portal solutions can significantly increase accuracy: firstly, the digital transfer of documents as well the integration with supplier scheduling tools allow for a greater transparency and control, so that processes can be optimised more easily. Cross-channel orchestration between repair service providers and incumbents can ensure that each transaction is mutually-agreed, appropriate, well-timed, and consistent. Secondly, the digitisation of invoices improves the understanding of cost structures and the speed of invoice validation. Real-time insights of raw material and service costs can assist insurers to better monitor and control payment sum required. For example, the German company ControlExpert offers not only a portal solution to connect with repairers and dealerships, but also data bases of invoices, fees, and costs of spare parts in order to help insurers better assess invoices. Better control of costs in the fulfilment stage can, from our experience, reduce claims leakage by a further 10%.
Examples of suppliers include: ControlExpert; Guidewire; Eucon
SETTLEMENT/CLOSURE – reduce litigation
As improvement of claims leakage protects customers against underpayment, it will in turn result in a reduction of litigation potential and by that reduce the need for legal reserves.
The challenge remains to digitise claims information and to start applying intelligence to the process of claims assessment and fraud detection. Even a small improvement in the control of claims indemnity expenses can have a significant impact on an insurer’s end result. From our zeb industry research, the void between As-Is models and indemnity saving potentials is far too great to ignore – rendering the need to review claims procedures and assessing them for optimisation potentials paramount for future profitability.
[i] Insurance Europe, European Motor Insurance Markets, 2015; Insurance Europe, European Motor Insurance Markets Addendum, 2016; Milliman, Driving for Profit, A view of the UK private and commercial motor insurance markets 2015, 2016; Fédération Francaise de l’Assurance, Chiffres clés, 2014
[ii] Association of British Insurers, Insurance Fraud Taskforce, 2016