On a daily basis, p&c insurance providers mine through vast repositories of data to validate and process thousands of claims.

Yet, billions of dollars are lost annually because of fraudulent insurance claims. In order to provide quality services to their customers, providers need to recover this lost money. Preventing fraud requires mining and analyzing massive volumes of data to gain better insights and, in turn, improve decision-making ability.

According to a recent survey by FICO¹ and Property Casualty Insurers Association of America (PCI), 45 percent of insurers estimated that insurance fraud costs represent 5 to 10 percent of their claims volume, while 32 percent said the ratio is as high as 20 percent. More than half (54 percent) of insurers expect to see an increase in the cost of fraud.

Now let's evaluate some of the challenges that persist and the changing insurance landscape that is driving innovative solutions and approaches.

Challenges include:

Information overload and the rise in the number of security threats and frauds.
Technological limitations that make it challenging to process and analyze data in a timely manner.
Information silos, disparate systems and departmental processes that create information leakages.
Evolving demand to keep up with changing compliance and regulations requirements.
Lack of skilled resources to investigate and address fraudulent activities.

A Changing Landscape Leads To New Realizations

While the insurance industry has matured over the past few years, traditional methods of fraud detection are unable to keep pace with the rapid advances in technology. Criminals these are days are sophisticated, constantly change their tactics and are very skilled in identifying the loop holes. As such, carriers need to implement a robust strategy built on advanced analytics foundation that is capable of handling security issues that arise because of information siloes as well detect fraud promptly.

Fraud can occur at any stage of the claims process, leading to a security breach, which is one of the biggest concerns for both for consumers as well as the insurance companies. The results of the Javelin Strategy & Research study, “Identity Fraud Report: Consumers Taking Control to Reduce their Risk of Fraud” revealed the number of people affected by data breaches has grown 67 percent since 2010.

In 2011, more than 11.6 million adults in the United States were victims of identity fraud and the number is increasing year on year. Organizations must start analyzing national and public security trends holistically across the organization to prevent large-scale threats. Using robust advanced analytics, organizations can accomplish the following goals:

Advance the precision of fraud detection.
Reduce false positive ratios.
Detect fraudulent claims before payment at a faster rate.
Reduce financial liability.
Reduce operational cost by using a common technology platform for fraud and security issues.

Organizations can also deploy these measures to counter fraud:

Connect the enterprise tightly. Deploy a common infrastructure across the organization and ensure a smooth flow of information across various systems to make it easy to analyze data from across channels such as users or accounts and provide a comprehensive view of an individual's relationship with the organization.
Monitor continuously. Leveraging advanced analytics to monitor and authorize transactions in real-time enables proactive identification of a fraudulent transaction with no negative effect on the customer experience, thereby protecting brand reputation.
Discover relationships in your data. Establish links between your data entities such as customers, products, accounts or services to easily identify organized or collaborative fraud activities that would otherwise go unnoticed.
Mash up structured and unstructured data. Useful information can often reside in unstructured data formats like claims log, survey reports, emails, social media and geospatial information. Once mashed up with the structured information from internal systems, this unstructured data can provide valuable insights into detecting claim patterns.

No single analytics approach is perfect and each organization needs to experiment and evolve to come up with a fraud detection system that works the best for them. That said, it is important to incorporate a hybrid analytics approach which combines various methods including monitoring past patterns and business rules, detecting anomalies, predictive data mining, analyzing social, champion-challenger adaptive segmentation with advanced neural networks, and relationship mapping for greater accuracy and improved detection.

Secondly, as mentioned earlier, combining information from structured data with the unstructured data is important to link unrelated events, unearth hidden facets and detect security threats and fraud in the early phases.

Footnote

¹FICO PCI Insurance Fraud Survey, FICO, October 4, 2012.

Special Reports /

Data Mining and Relationship Mapping: Hybrid Approaches To Fraud Detection

Related Stories

Recommended For You