The Art of Identifying Scams

Claims departments generally face two main challenges: providing optimal customer service and accurately detecting and mitigating fraud. While some may not see a distinct link between the two, there is a common thread that could provide carriers with a solution to both challenges. 

Ironically, fraud detection processes are not seen as particularly customer-friendly. Current approaches to fraud detection trigger excessive false-positive red flags, slowing the claims process and frustrating good customers. Fraud fighting also is largely a manual, resource-intensive endeavor handled methodically by seasoned claims professionals whose skill sets are increasingly in short supply.  

Automated fraud detection is the common thread that links both fraud mitigation and improved customer service. Smart carriers are recognizing that fraud protection is not just about flagging claimants who shouldn’t be paid; it’s about instantly identifying those who should be paid so carriers can resolve cases quickly and deliver a positive customer experience. 

The key is to use the right fraud detection tools early and throughout the claims process. Adding robust data and analytics starting at the first notice of loss (FNOL) and throughout the claims resolution cycle can cut time, minimize costs, and forge stronger bonds at the claimant’s moment of need. Let’s dig deeper into how data and analytics technologies can be applied to each challenge.

Improving the Customer Experience

Competition in the insurance market requires carriers to be cost efficient and to provide their customers with a better experience in order to keep them happy and loyal. This dynamic places claims professionals between two opposing forces. First, they are under pressure to lower their loss adjustment expenses in order to improve competitiveness. At the same time, they are asked to do this in a way that increases customer satisfaction, since customers can be retained or lost during the claims process. 

According to a J.D. Power and Associates satisfaction survey on auto claims, “Among delighted claimants (overall satisfaction of 900 or higher), 83 percent say they ‘definitely will’ renew their policies and 84 percent ‘definitely will’ recommend their insurer. Among displeased claimants (scores of 549 and below), only 10 percent say they ‘definitely will’ renew and 10 percent ‘definitely will’ recommend their current insurer.”

Identifying fraudulent claims faster and earlier in the claims process can address both competitive pressures. FNOL is a good starting point, since it often is the only interaction a customer has with the carrier following the quote or most recent renewal. The faster a fraudulent claim can be flagged while good claims are expedited, the more competitive the carrier can be. Similarly, the more confident a carrier can be in a claim, the sooner it can be fast-tracked. 

Claims data prefill technology is one way to establish confidence early in the claims process. It can instantly populate all or most customer data fields from a single piece of customer data (such as a phone number). Data prefill improves fraud detection by giving carriers the most accurate and up-to-date data to spot suspicious claims. It reduces data entry errors, call times, subrogation referrals, and uninsured motorist identification times. Nonfraudulent customers get a hassle-free claims experience, minimal disruption to their daily lives, clear communication throughout the process, and timely resolution of their claims.

Continuous Evaluation

Fraud mitigation does not start and end at the FNOL stage. Carriers should employ automation technologies that continuously evaluate the claim until it is closed. Claims are constantly evolving and acquiring new information throughout their lifecycle. Information that might identify possible fraud can appear at any time. A good, automatic, fraud-detection solution never sleeps. Instead, it continuously monitors the claim on every update for potential red flags. 

It also gives carriers another layer of confidence at a time when the insurance industry is beginning to experience a skill-set shortage in the claims department. Seasoned claims staff are aging out. As tenured pros exit the workforce, the younger, less-skilled staff identifying claims will inevitably result in a drop in the quantity or quality of fraudulent claims referrals. This inexperience can lead to increased operating costs as well as to lower customer satisfaction due to delays in claims handling.   

Automated fraud detection can help address this inexperience gap through its ability to analyze both structured and unstructured data. Structured data—such as name, address, and Social Security number—is routine. But estimates suggest that 85 percent to 97 percent of all the important data in a claim is unstructured, mostly within the file notes. This unstructured data often contains terms, phrases, and even code phrases that could be of value in spotting fraud. Phrases like “claimant uncooperative” and “brake lights not working,” or conflicting accounts of how many passengers were in a vehicle are potential fraud indicators. Taken out of context, these phrases could produce a large number of false positives. Good automation technology can identify and feed the appropriate information to the claims professional or investigator. 

Predictive Analytics

Insurance companies have been using analytics on the underwriting side of the business for years but little on the claims side. With the growing volume of data retained with new claims systems, it is more important than ever to use the data to optimize claims processing and improve fraud detection capabilities. This is where predictive analytics can play a key role in automating fraud detection. 

Predictive analytics divide a carrier’s existing data sets to develop models that identify patterns and predict future potential outcomes. Think of it as an early warning and alert system that can notify carriers of claims that are potentially fraudulent so that they can take action earlier. 

Predictive analytics can flag medical providers with suspicious billing activity or unusual treatment patterns and identify policies with discrepancies or irregular combinations. The technology also can automate and standardize traditional, manual, fraud-detection processes, providing a new level of efficiency when spotting potential fraud.

Additionally, rules-based workflows can spot trends or anomalies automatically that manual processes might overlook. Extremely advanced forms of analytics, such as neuro-network models or machine learning, can even use predictive analytics to alert users about suspicious transactions in real time. Again, all this efficiency means less time is spent chasing fraud while more time and resources can be spent enhancing the good customer’s experience.

Internal and External Data

Another area where carriers can improve the customer experience while mitigating fraud is by integrating external data with their internal data. Today, most carriers work with only the claims data within their domain. Pulling in data from public records and third-party data sources can shine a whole new light on a claimant’s information. 

Some examples of public records and other external information that could assist in identifying and resolving fraud include:  

• A late filed/reported bankruptcy—one that could not be seen immediately after FNOL or when the claims professional completed their initial background check.
• Prior felony convictions.
• Potential fraudulent activity of a noninsurance nature.
• Multiple Social Security numbers attached to a single party, connected to multiple parties, connected to a deceased person, or issued prior to a date of birth.
• Liens and lawsuits.
• Numerous insurance policies issued in the last 12 months.
• Multiple claims with similar facts of loss over the last 12 to 26 months.

Cross-industry databases also can be helpful in flagging claims because fraud is often part of a pattern of bad behavior. Research has found that government or retail fraudsters are more likely to commit fraud in another industry, such as property and casualty insurance. 

For example, a Georgia resident’s claim and policy ID are unlikely to emerge in an Ohio claim. The ability to confirm this type of information in real time would be a tremendous asset for flagging a suspicious claim and alerting the investigative team.

Watch-list information on known fraudsters also can be incorporated into automated fraud detection platforms to continuously monitor and alert carriers when a known individual is in their system. Even a common fraud detection asset such as a police report can be automated. Technology now exists to automate the ordering and delivery of records in near real time. In some jurisdictions, carriers can access police record data faster than they ever could before, leading to speedier investigations and ultimately claims resolution. When you consider having claims data, public records, watch-list data, and cross-industry data feeding a fraud-detection system, imagine how much clearer a picture would emerge of each claimant.

Competition has made it critical for carriers to deliver an exceptional customer experience, especially during the claims process. Yet the fraud detection that impacts claims resolution has remained a primarily manual undertaking. The good news is that claims departments are recognizing the need for automation. This shift is driven by the need for greater process efficiency, a push for lower loss adjustment expense, and a desire to enhance customer service by increasing ease of use for consumers.

By utilizing more data and analytics to automate fraud detection from the beginning and throughout the claims life cycle, a carrier can gain greater confidence in fast-tracking claims, which leads to a customer experience that can truly differentiate it from the competition.