Fraud occurs when someone knowingly lies to obtain some benefit to which they are not otherwise entitled to or someone knowingly denies some benefit that is due and to which someone is entitled. Insurance fraud occurs on both the claims and underwriting side, but we will limit the focus of this post to claims fraud. Claims fraud can be broken into two categories – “Soft Fraud” and “Hard Fraud”. Soft Fraud is used to define unwanted opportunistic behavior of normally honest people. Hard Fraud represents carefully premeditated and executed scams to stage or invent an accident, injury, theft or other incident to benefit from an insurance claims payout.
With thirty billion dollars lost every year to P/C claims fraud in the U.S. alone, I was interested in better understanding the industry landscape in terms of how claims fraud is fought by carriers today. After a series of discussions with several insurance carriers, here is an attempt to quantify the industry landscape in the form of a maturity model based on capability tiers. For the sake of simplicity, I’ll classify insurance carriers into five tiers – Levels 1 to 5, with 5 being the best in its ability to fight fraud. The four dimensions that are used for this assessment are: (1.) analytic tools to detect soft and hard fraud; (2.) effective use of data in the investigation progress; (3.) level of workflow integration and automation; and (4.) Special Investigative Unit (SIU) department personnel’s analytical capability. The main metric that is used to determine the tier in which a carrier would fall into is called the Actionable Rate of Fraud Mitigation. The Actionable Rate is defined as the percentage of the total referred suspicious claims that identify a confirmed case of fraud where a mitigation action can be taken. An optimal referral rate and a higher actionable rate are desired to effectively mitigate fraud.
Please view this as a flexible broad-brush framework to get a general sense for how claims fraud is being fought by the industry today and the optimal level of analytical capabilities that are required to effectively fight fraud. It is impossible to capture the numerous permutations and combinations of the various components that are part of a carrier’s fraud-fighting weaponry and classify them into rigid tiers.
Level 1: Carriers in this tier predominantly depend on their claims adjustors to provide a majority of the referrals. Instead of using analytical tools, SIU departments and claims personnel rely on their years of experience to detect fraud. Carriers mainly depend on the information available in the claim files and perform manual lookups against industry watch lists on suspicious records. There is very limited claims workflow automation and no ability to proactively stop payments on suspicious claims. It is common to find offline excel spreadsheets being used to track investigations as opposed to industry-standard case management systems. The SIU department size and budgets are significantly constrained and SIU personnel are not adequately trained on the latest analytic techniques. Carriers in this tier suffer from a low SIU referral rate and a significant percentage of fraudulent claims go undetected, leading to a very low actionable rate.
Level 2: Carriers in this tier don’t differ much from the previous tier, except for their use of business rules. In most cases, the homegrown set of business rules and commonly-used industry guidelines are manually applied during the claims review process. Claims that violate the business rules, or appear suspicious based on manual industry watch list lookups, are referred to the SIU department. These carriers also get a majority of their referrals from adjustors and have very limited automation and ability to proactively stop payments. The actionable rate is slightly better than the previous tier, but still sub- optimal.
Level 3: Carriers in this tier are seen using generic predictive models along with business rules to detect fraud. The biggest challenge faced by some of the carriers here is the high rate of false positives driven by their predictive models, which impacts customer experience and forces them to abandon the models in due course. Carriers in this level use LOB-specific industry datasets and manual industry watch list lookups in the investigation process. Usually the business rules are well-automated and the payment workflow is somewhat automated.
Level 4: At this level, carriers use a combination of business rules, generic predictive models, and canned network link analysis reports from external sources to detect fraud. The business rules lookups are well-automated and the payment workflow is somewhat automated. Most of them use industry-standard case management platforms, but the fraud workflow is mostly disconnected from the core claims management system. Carriers still lack the ability to access all the data files in one location to conduct a holistic analytics review. The SIU departments are well-staffed and somewhat analytical. There is usually at least limited support from an in-house or external Data Science team for analytics.
Level 5: Carriers that operate at this level enjoy the highest actionable rate and a significant bottom-line savings from their ability to effectively mitigate fraud. The SIU departments are well-staffed, well-funded, and very analytical. The carriers use a combination of business rules, LOB-specific and fraud-type specific predictive models, as well as robust social network or network link analysis driven by big data platforms to detect fraud. The relationships that can be picked up from graph databases can be valuable in detecting hard fraud committed by organized fraud rings. Rather than using predictive models generated from generic historical claims data, carriers at this level use LOB-specific industry datasets for specific types of fraud to train the best models (e.g., contents industry data to develop a contents theft fraud model). The models found here are optimized for false positives and are purpose-built. The business rules lookups and payment workflow is fully automated to stop or delay payments at FNOL on suspicious claims coming from the regular and digital channels. Most of the carriers use industry-standard case management platforms and the fraud workflow is fully integrated with the core claims management system. Carriers have the ability to access all the data files (FNOL audio, images, video, case files, external text documents, etc.) in one location to conduct a holistic analytics review. Carriers in this tier are able to find the right balance between detecting fraudulent claims and delivering a superior claims settlement customer experience by fast-forwarding the legitimate claims for payment.
Optimizing claims fraud protection presents a great opportunity for carriers to shave a few points off their combined ratio. This mundane standard business function can be transformed into a source of competitive advantage with the right solution framework. Advanced analytic tools and workflow automation are just part of an effective fraud mitigation framework, and these tools by themselves don’t create a game changer. It takes the right combination of talented SIU personnel with years of experience, advanced analytic capabilities, and smart workflow integration to create that magical force that is required to effectively fight fraud. The majority of the industry today operates in Levels 2-4, but we will be seeing more carriers striving to achieve actionable rate at a Level 5 maturity in the coming years.