A lot has been written about criminal activity resulting in losses due to P&C fraud. While coverage of the actual crimes is pervasive in the media, not enough has been written about strategies to prevent these losses. Let's explore how recent innovations in analytic technologies can help prevent fraud and contain escalating claims.
The Challenge
Today more than ever, the volume and velocity of incoming data coupled with shrinking windows of time make fraud investigative analytics very challenging. Recently, we spoke with a senior P&C investigator at a large insurance carrier, who pointed out a series of challenges with existing technology used in investigative analytics today.
The sophistication and complexity of the techniques employed by fraudsters have increased. Offenders have been highly effective at avoiding detection and hiding in the massive amounts of claim data. The investigator noted that the ability to track new schemes across personal and commercial lines of business would be valuable, but that unified approaches spanning disparate data sources have been extremely difficult to achieve. Analyzing data at different phases in the insurance process can point out anomalies that warrant additional investigation before claims are paid.
He also stated that existing technology focused on investigative analytics was not designed with the investigator in mind. It either focuses on automated detection algorithms or constrains the investigation by not allowing enough freedom to explore a variety of data sources. Providing an intuitive and usable analytical environment for the investigator to apply his training has been an elusive goal.
Over the past few years, most of the innovation in analytics has been in the area of automated information analysis. These techniques remove the analyst from the equation by attempting to reveal all relevant insights automatically. However, in most investigations, the single most important component is human judgment.
Where Is the Innovation?
Fortunately, modern technology has caught up with these challenges. Three breakthroughs in analytic technology can be applied to support your fraud investigation techniques. They are:
- Interactive data visualization;
- Collaborative investigative analysis; and
- Unified data views.
These three innovations attack the problem at its core and expose what the criminal wants to hide most. Together, they represent the foundation of a new approach in fraud investigative analysis called Interactive Analytics (IA). This approach has been used by the U.S. government to address some of the most demanding applications in the world, including counter-terrorism, cyber crime, and financial investigations.
IA is an investigative-centric approach to analyzing and understanding data in support of more accurate identification. This approach leads to improvements in detection, reporting, and case resolution. IA provides the ability to explore, detect, and confirm hidden relationships across disparate data sets. The exploration can be conducted in an unconstrained manner while investigators exchange insights as they are uncovered.
How do Interactive Analytics work? In a very simplified form, IA consists of connecting to valuable sources of data for analysis, freely exploring the data to identify fraud, and sharing insights with other investigators. Here is a brief example of the three modern breakthroughs in analytics in the context of P&C fraud.
Challenge: Fraudsters are good at avoiding detection.
Innovation: Interactive Data Visualization
Interactive data visualization allows the investigator to represent massive amounts of data in different visual representations and, in so doing, isolate important facts and patterns. But it is more than just pictures. It gives the investigator the freedom to ask questions through direct interaction with the visualizations. As the investigator discovers new insights, he can easily navigate, drill down, and develop vivid profiles. He can link together individuals, suspicious events, past claims, current and historical alerts, accounts, acquaintances, background checks, and transactional behavior. This is a powerful approach to investigative analysis that leverages the ultimate pattern recognition machine: the mind of the investigator. The ability to detect non-obvious relationships and confirm inferences very quickly are important characteristics of interactive data visualization.
In Figure 1, insurance customers are linked to Property Fire Claims and Property Theft Claims (alert types). These alert types are highlighted in orange. Focusing on Figure 2, one customer (Jim Baker) has three claim alerts: a home destroyed by fire, a multiple durable items claim, and a claim for a luxury item. Many other customers also have multiple claims across insurance alert types.
Interactive data visualization allows fraud analysts to visualize this data in any form. The examples shown happen to be link analysis visualizations. The same data can be presented in other forms, including charts, heat-maps, geospatial views, tables, and event timelines. By shifting their lens from one visualization to another, investigators can quickly identify suspicious claims leading to more in-depth and accurate investigations.
Challenge: Investigators need to share insights and work together.
Innovation: Collaborative Analysis
Investigative analysis is a lonely function in most organizations. Even in some of the most well known financial institutions, business lines and investigative groups assigned to those business lines are separate. Interactive Analytics allow investigators to perform collaborative analysis. When insights are revealed, they can be instantly shared with others. In addition to viewing the final artifact or graph, collaborators gain visibility into the entire analytic process or session that led to the conclusion. This can be extremely useful towards achieving the elusive goal of group-based analysis or collectively following a line of reasoning. The result is an investigative organization that can actually leverage the collective domain knowledge of its employees to improve efficiency and accurately identify fraudulent activity.
Exchanging insights brings many more resources to bear while also recording a history of the analysis, which can be used in subsequent litigation support. Training is another benefit of collaborative analysis. It is a useful way to coach new investigators and review the steps in the investigative analytical process.
Challenge: Disconnected data in too many places.
Solution: Unified Data Views
Without a complete picture of the investigative landscape, it is difficult to solve cases. The ability to unify data into a single view can be one of the most effective ways to further investigations. In this case, the analyst may want to bring in account detail, insurance agent names, related people, or other important facts. These valuable pieces of information may exist in different systems, making this problem more complex. Unified views overcome this challenge by allowing the investigator to incrementally evolve the picture by bringing in additional data from new sources as it becomes available. The result is a more complete profile.
Experts say that in difficult economic times, fraud schemes are on the rise. At a time when the volume and velocity of data is increasing at an alarming rate and investigators are given less time to get the job done, new approaches are needed. Interactive Analytics brings together three innovations in analysis to increase the chances of solving fraud cases in a timely fashion. It facilitates identification, collaboration within and across investigation units, and unified data to create a vivid portrait of what is really going on. Solutions like this can be deployed quickly in support of real organizational goals.
© Arc, All Rights Reserved. Request academic re-use from www.copyright.com. All other uses, submit a request to TMSalesOperations@arc-network.com. For more information visit Asset & Logo Licensing.