Insurance fraud fighting and risk assessment best practices series. What we’ve learned from over 150 implementations at P&C insurance carriers globally.
Data scientists at carriers deal with huge amounts of information: internal insurance data from files or coworkers, data from various systems and data from external sources. Information on the insured persons and assets, claims and detected fraud helps in making well-founded judgements about risks, trends and the value of policies and portfolios. In an ideal world information would be captured completely in figures and data fields that make sense. But how reliable is all this information? There are substantial pitfalls between the ideal world and reality: both in the systems and with us humans. Differences in culture, accuracy and consistency make it difficult to compare the contents of multiple administrative systems. And to top it off, the human factor can have both a positive and a negative influence on data quality.
Increased focus on data quality
The quality of a carrier’s data has become increasingly important. There is a growing industry awareness that quality information is essential to improving the customer experience. Making use of good data ensures short acceptance and claim processing times resulting in happier customers. This data must be readily available and consistently reliable.
In our recent Insurance Fraud Survey, 45% of insurers report a challenge with the quality of fraud data collected. In 2016, only 30% of insurers saw this as a major challenge. The main reasons noted were that too little information was available and/or poor-quality information disrupted the process of effective and reliable analysis.
Predictive models, network analysis, text mining and machine learning techniques could really help insurers to assess risks in underwriting, claims and during special investigations better and faster. Carriers need to begin their AI endeavors with the existing proprietary data, and subsequently enrich it with third party data, to get actionable insights and perform machine learning. The end result is an automated fraud monitoring model that triggers a self-learning process. Christian van Leeuwen, FRISS co-founder and CTO: “We restore the win-win situation for both the parties while helping insurers fight fraudsters. We firmly believe that the FRISS Score should be a “white box,” delivering actionable insights using techniques such as visualization.”
Exchange of authorized information between carriers
Today’s insurance customers are more apt to request quotes and purchase insurance online. It’s easier and more convenient, and many companies are now actively encouraging it over traditional phone or in-office visits. Focusing on online interactions makes it more important for insurance companies to have immediate access to high-quality data to make smart decisions on who to insure. The problem is that most companies don’t have access to enough of it.
Pooling authorized data helps. Data shared can provide hit/no-hit signs to incoming applications, claims or renewals. Shared data includes information about false claims, unreliable repair shops and health professionals, imagery and information about insured assets. Years ago, this was virtually impossible. Fraud data pools allow insurers to detect and prevent fraud quicker and more accurately – a benefit to all companies while remaining fully compliant, as no information is shared other than a green or red signal.
Access to external data
Insurers would be able to fight fraud more effectively if access to external databases would be easier. Information from external sources may present a more comprehensive picture, thus providing good and sound arguments for acceptance, rejection or perhaps adapted conditions.
External sources can include information from a chamber of commerce, payment behavior and claims history, as well as demographic data and vehicle records.
Setting the scene
No insurance company is immune to fraud. As much as the industry might feel prepared, fraudsters are smart and always look for the weak spot. Fraudsters use everything they have in their power to get money from insurers and they find ways to avoid getting caught. And these fraudsters are insurer-agnostic, so no one is safe. Insurance fraud is a global problem. On average, 10% of incurred losses are related to fraud. Fraud is also a growing problem, contributing to 10 to 15 percent of total claims costs. The total cost of P&C insurance fraud is more than US$80 billion per year in the US alone, according to the Coalition Against Insurance Fraud. That means insurance fraud costs the average US family between $400 and $700 per year in the form of increased premiums.
By actively fighting fraud we can improve these ratios and the customer experience at the same time. It’s time to take our anti-fraud efforts to a higher level.
The good news here is that the battle against fraud is at least being taken more seriously. Fraud affects the entire industry, and fighting it pays off. US insurers say that fraud has climbed over 60% over the last three years. Meanwhile, the total savings of proven fraud cases exceeded $116 million. Insurers are seeing an increase in fraudulent cases and believe awareness and cooperation between departments is key to stopping this costly problem. The insurance industry is working hard to improve on fraud detection and prevention. It is definitely a topic on the agenda and not underestimated. In this case, everything starts with awareness.