The Rise of Big Data in Detecting Insurance Fraud

Published Aug-13-18

Using big data, prediction modeling and machine learning to detect fraudulent insurance claims.

Insurance industry, United States

The Story:

The Rise of Big Data in Detecting Insurance Fraud Insurance fraud is a huge problem that costs insurers billions of dollars. In testimony to the USA Congress in 2017, the head of the Insurance Information Institute estimated that the cost to the industry of property, casualty and auto insurance fraud is $30 billion per annum. Some estimates state that 25% of all claims contain an element of deception. Fraud also costs individuals. The FBI’s insurance fraud statistics state that policyholders are paying between $400 and $700 more on their premiums because of fraud.

Consequently, companies are keen to use and invest in the latest technologies to spot and catch the fraudsters before they take even more dollars. And one of the areas they're focusing on is big data.

Vast Reams of Customer Data

Organizations are now relying on a wealth of data that they hold about customers such as analyzing call center conversations and notes from customer service agents alongside social media accounts and data held by third parties. This can include address changes, criminal records and people’s bills.

Manually checking all of this information would be time-consuming, expensive and nigh on impossible. However, predictive modeling can detect fraud by identifying patterns and relationships between people and data. A simple example of how this works is if a person claims their basement was flooded during a storm on a particular day. This can be cross-checked with weather reports and their social media accounts. If the claimant has posted photographs of them enjoying a barbecue in their backyard on the day in question and the sun is out in full it could indicate that they’re being less than truthful with their claim.

Social network analysis (SNA) would also be able to pick up whether an address given by an accident victim has been used for multiple claims or if a car has been involved in several crash incidents that resulted in claims.

Insurance companies can also rely on text analytics technology to see if customers are making false statements. Fraudsters tend to alter their stories with repeated telling often in very subtle ways and these can be picked up by the technology.


The use of big data is helping in other ways too such as with telematics-based solutions to spot and prevent fraudulent car insurance claims. One organization Octo Telematics uses Internet of Things (IoT) strategies to analyze billions of data points from connected vehicles. A black box is fitted to an individual’s car which records such information as driving speed, time of drive and GPS location.

An example of this can be useful is with whiplash injury claims which experienced a dramatic upsurge in recent times as consumers were encouraged to pursue claims by 'ambulance-chasing' lawyers. Telematics technology can provide crash reconstruction dossiers that go into forensic detail and from them, it can be ascertained whether a vehicle was traveling at a sufficient speed to cause a whiplash injury.

Outwitting the Fraudsters

Aware of the potential of big data analytics to save the industry billions of dollars, an increasing number of companies are looking to improve their IT infrastructure to stay several steps ahead of the fraudsters.

Share on      
Next Story »

What Can we Solve for You?