Data analysts often use the term "big data" to describe data sets that are too large to store and analyze with traditional methods such as a relational database management systems (RDBMS). Big data is greatly affecting the operations of many businesses, including those in the insurance sector. Insurance data can be particularly challenging to use because it comes from many sources such as adjusters' notes, fraud lists and claims databases.

The large number of claims that insurance companies receive means that adjusters often fail to review all the available information, which can result in a poor decision. The role of data analytics is becoming increasingly important in insurance for tasks such as identifying claims for closer inspection and prioritizing claims. Even a slight improvement in an insurance company's loss ratio can mean a great improvement on the bottom line, especially for large insurers.

The following six areas show where big data is having the greatest impact on the insurance sector:

  • Settlement

  • Loss reserve

  • Fraud

  • Activity

  • Litigation

  • Subrogation

Settlement

Settling claims quickly is one of the highest priorities for an insurance company. Most insurers have implemented fast-track processes for claims that have a low risk of being fraudulent. However, this process can result in an insurer paying for claims it shouldn't. This possibility is especially likely when an insurer receives many claims at once from the same geographic location, which occurs during natural disasters.

Big data can help analyze claims and the claimant's history to optimize the number of claims eligible for instant payouts. Analytics can also reduce the time needed for adjusters to manually evaluate claims. Additional benefits of processing claims quickly include savings on rental cars while the claimant is waiting for a payout.

Loss Reserve

Loss reserve is an insurer's liability from future claims. Insurers need to estimate their loss reserve accurately in order to underwrite as many policies as possible without allowing liabilities to exceed assets.

It's generally impractical for an insurer to predict the payout and processing time when a claimant first reports it. However, accurate claims forecasting is particularly important for potentially large claims such as workers' compensation and liability. Big data analytics can help with estimating loss reserves by comparing a new claim with processed claims similar to it. This software can then reassess the company's loss reserve each time the claim data is updated, so the insurer always knows how much money it needs to keep on hand to meet future claims.

Fraud

About 10 percent of all insurance claims are fraudulent, making it critical for an insurance company to detect as many of them as possible before making a payout. The traditional method approach to detecting fraud is to create a set of fixed rules that determine when to pay a claim. However, criminals are able to determine these rules with relative ease and figure out ways to get around them.

Predictive analysis uses rules as well, but it also uses data mining, modeling and exception handling to make it much more difficult for insurance fraud to succeed. Analytics also detects fraud earlier in the claims cycle than rule-based detection.

Activity

Insurers providing cheap car insurance generally want to assign the most complex claims to their most experienced adjusters. However, it's often difficult to determine a claim's complexity when the claimant initially reports it. Insurers often assign a case to a new adjuster at first because the case seems simple, but end up reassigning it to a veteran adjuster when the case turns out to be more complicated than it initially appeared. Reassignments increase the time needed to process a claim, thus reducing customer satisfaction.

Analytics can use data mining and group loss characteristics to score claims on complexity and assign them based on the adjuster's experience level. It may also be able to adjudicate and settle claims automatically.

Litigation

Insurers spend a significant portion of their loss adjustment expenses in defending claim denials. Analytics can calculate the litigation propensity for a claim, allowing insurers to assign claims with a greater likelihood for litigation to a senior adjuster. This strategy helps insurers settle claims more quickly and for smaller amounts.

Subrogation

Subrogation cases occur when a third party is responsible for paying off an insurance claim. Detecting these cases early in the claims process reduces loss expenses and increases recovery rates. The large volume of data needed to identify these cases often means they go unidentified, but text searches can locate phrases in this unstructured data that could indicate a subrogation case.

Summary

Insurers are making the analysis of big data part of their regular claims process. This capability is especially useful in insurance due to the many sources of unstructured data in this industry. The potential for reducing the cost of each claim can result in a large savings for insurers and their policyholders.