Operational Excellence & Technology
A look at how one carrier achieves success in fraud prevention.
Addressing operational excellence and technology at once can be intimidating for many seasoned managers. However, let’s begin with the understanding that this will not be a deeply technical article. Here, my hope is to put forth experience and observations that will stimulate discussion. Perhaps carriers will be prompted to step back and take a hard look at their fraud prevention programs to determine whether or not they’ve reached their potential or can improve upon their results by implementing technology.
What distinguishes one program from another and who’s to say what true operational excellence is? Many would say that this is a subjective goal. As an industry, we’ve struggled to benchmark programs since no two companies are the same. Whether due to distribution sources, underwriting guidelines, geography, or company philosophy, we remain challenged to achieve a consensus on these types of issues.
Planning and Determination
Operational excellence is something that’s determined by an individual or company and will vary based on numerous factors. You don’t reach operational excellence without careful planning and determination. And, factors outside of your control may hinder your ability to reach certain levels of operational excellence. However, proper planning will take these factors into account and your plan can be built around them. You must optimize what you have available to achieve a level of performance that your resources and budget allow. Establishing a clear vision and action plan is essential to this initiative. As managers, we work within a defined framework every day. Knowing our limitations will allow us to plan accordingly. For instance, if budgetary constraints prevent you from hiring additional investigators, your plan to achieve OE cannot include new hires.
The company I work for sees the value that a strong anti-fraud program brings to the bottom line and its reputation. Having worked in a special investigations unit for nearly 17 years, I’ve seen a lot of changes in this industry. Some of the changes have been good, others not as good. Those of us that have been in the insurance fraud prevention/investigation business for 10 or more years have seen the increased utilization of technology in our day-to-day work. Nothing has been more dramatic in our business than our reliance on data and technology to improve upon the core skills many of us brought to this profession.
We recognized early on that while there are many factors within a program, technology would be needed in our effort to attain operational excellence (OE). Our program has evolved over the years, and we see technology as an area that constantly will change.
In 2001, we realized the need to embed technology into our program. Our initial venture was from a data mining standpoint. We were able to identify abnormal claim patterns associated with either rate evasion or large scale theft ring activity. While we were pleased with the results, the process was manual and most abnormalities were found after the fact. To achieve OE from a proactive standpoint, we knew we needed to catch things much sooner in the life of the claim.
We entered into an agreement with Computer Science Corporation (CSC) to co-develop a new product that could meet our expectations. During 2002, subject matter experts from the SIU, along with various MetLife technology units, worked with a team from CSC to develop a product which was placed into production in February 2003. The premise was to develop an application that would identify potential indicators of fraud at the first notice of loss (FNOL) and provide a score for review to determine whether or not a formal referral to the SIU was warranted.
Fraud Evaluator ™, as it is known within the market place, is a claims fraud detection system that continuously checks claims data for suspicious factors, beginning with the FNOL and continuing through the life of the claim. The system uses claim data as it is captured by the company’s claims system, requiring no additional work by the claims handler. Each claim is scanned for various fraud indicators, or more sophisticated profiles, and relevant internal and/or external databases are searched for links to prior activity.
Claims are initially scored at the FNOL and are continuously re-scored as new data elements are captured. When the aggregate score reaches a user-defined threshold, the claim is “referred” for further review and possible investigation.
Some of the key benefits we have seen from this initiative include:
- Better customer service
- Early detection of possible fraud—before the claim is paid
- Constant screening of all claims
- Consistent application of detection rules
- Automatic referrals and e-mail notifications
- Consolidated access to 3rd party data sources
- Flexible configuration of scoring modules
- Internet/Intranet browser-based access
- Speed in processing of legitimate claims
The product has three unique technologies that we utilize to maximize our analysis of claim data. The first is a Predictive Model Engine (PME). When the product initially was built, we utilized a modeling company to provide the technology. As our skill level increased, we were able to partner with an internal group at MetLife that had the technology and ability to create models. We’ve since developed several predictive models which we plug into Fraud Evaluator. This allows us to be timelier in refreshing models and controlling the process to ensure all of our criteria are met.
The second key component to Fraud Evaluator ™ is the Identity Search Engine (ISE). This engine has a fuzzy search capability which allows us to scrub our own claim data against several external data sources such as commercial mail addresses, sanctioned medical providers, the National Insurance Crime Bureau and our own internal watch list.
The third piece to the technology puzzle is the Business Rules Engine (BRE). Through our business knowledge, we have provided numerous business rules. A majority of these rules identify industry-wide indicators, while other rules that have been provided are unique to our business needs and lines of business. The flexibility of the tool allows us to weigh certain aspects of a claim. These weights may take into consideration geography and/or loss types, depending on coverage types and/or policy provisions.
We believe that Fraud Evaluator, and our approach to its utilization, is unique compared to other products we’ve seen in the industry.
Where do we go next? Well, one constant for us is that we’re always looking to see what’s around the corner and what new technology will improve our results. Recently, we started to use text mining on our homeowner claims data. We found, based on the coverage, that there were fewer data fields to analyze, so text mining adjuster notes would be the most productive in this area. Given the success we’ve seen, we have begun to expand text mining into unstructured auto data as well.
Operational excellence is something we constantly strive to achieve. From a technology standpoint, I’d like to think we’ve come close to achieving it. But newly introduced factors, such as business growth, budget changes, and technology, weigh heavily on the program and force constant reassessment. So whether or not operational excellence is truly achievable is another thing we can add to the list of immeasurables in the industry.
We haven’t covered all areas of operational excellence that could be discussed, but I hope this article leads to further discussion, planning and improved fraud prevention programs.
John Sargent is the director responsible for the Special Investigation Unit (SIU) of Metropolitan Property and Casualty Insurance Company (MetLife Auto & Home) and affiliates, a subsidiary of MetLife, Inc.
John Sargent is the director responsible for the Special Investigation Unit (SIU) of Metropolitan Property and Casualty Insurance Company(MetLife Auto & Home) and affiliates, a subsidiary of MetLife, Inc.