Organizations have been investing heavily in AI recently, and the insurance industry is no different.
Insurers are focusing AI investments on fraud detection, with predictive modeling being the primary method to fight claims fraud. In the first part of this blog series, I mentioned how predictive analytics in risk scoring is one of the key innovations in fraud detection.
But that innovation is only as good as the data on which it’s built. Let’s examine why bigger and better data is critical to effective fraud analytics.
Data fuels fraud detection
Data is essential to anti-fraud technology—both the quality and quantity of data. According to a recent survey, 64 percent of insurers said poor data quality is the most significant challenge to implementing anti-fraud technology.
It’s important to consider the kind of enriched data that’s incorporated into a risk scoring solution. There are interesting data sets available to help generate quality analytics. These include:
- Matching claim history data
- Bankruptcy and foreclosure records to help understand possible financial motives
- Civil and criminal records to understand an involved party’s background
- Weather data to verify the location and time of loss are consistent with the loss report and claim severity
- Vehicle sightings data to determine if vehicle and involved party location is accurate
- Image data sets to recognize prior loss images, duplicate images, and altered images
- Synthetic ID checks to uncover false identities and enhance entity resolution
- Policy insights to provide policy underwriting information
Of course, the amount of data matters greatly as well. The more data analyzed, the more effective the solution will be at identifying suspicious patterns and characteristics. That’s why one may want to consider how a solution combines or aggregates data.
When multiple carriers across the insurance industry add their data together, the data set becomes much larger and more powerful. Without aggregated data, a carrier is limited to its own data and potentially some small third-party data sets. But one carrier’s data, regardless of the company's size, may only be a small piece of the puzzle compared to the industry as a whole.
It’s much easier to recognize a suspicious loss with the benefit of matching claim history data across the entire industry, lines of business, coverage types, vehicles, and involved parties because fraudsters don’t limit their activity to one carrier.
Industrywide aggregated data is often the basis for the most effective predictive analytics because the bigger the data, the better the analytics.
The big data advantage
Let’s look at a few examples of how a fraud analytics solution incorporating industrywide data can make a significant difference:
- The solution could notify an investigator that an involved party submitted three prior water damage claims with three different insurers for the same property
- It could flag a claim by analyzing a vehicle’s salvage history and theft history regardless of who previously insured the vehicle or who’s submitting the current claim
- It could identify that an involved party has a penchant for filing wage loss, disability, and workers’ comp claims by taking out different policies with different carriers
These types of policies and claims, prior injuries and associations, and frequencies in claim and policy history are all relevant and provide some of the strongest fuel for planning and conducting investigations. The information can lead to resisting payment for prior or overlapping damage, detecting organized activity, or identifying falsifications and non-meritorious claims that could lead to either policy rescission or mitigation.
But that information may only be available when you have access to industrywide data.
Getting ROI from fraud analytics
Data drives advanced anti-fraud technology. But not just any data will do. You need quality, accurate data as well as broad, industrywide aggregated data for predictive analytics to be effective.
Insurers are making significant investments in AI-powered fraud detection. But unless those investments include the right data, they may not get the ROI they desire. The right data makes it easier to operationalize and adopt robust analytics. Verisk leverages industrywide claims and medical billing data sets to power its anti-fraud solutions. Our ClaimSearch data is aggregated from more than 90 percent of the P&C industry’s insurers and includes more than 1.6 billion claims. The Aggregated Medical Database sits on Verisk’s platform and contains billing data from 2 million medical providers.
And it’s easy to pilot our solutions—we use your current data and it requires no IT lift. Those assets, combined with innovative predictive analytics, help deliver better results in the fight against claims fraud.