In Few Years, Social Network Data May Be Used in Underwriting

From Insurance Journal by Young Ha

The insurance industry is paying increasing attention to what people and businesses post on social networking sites like Facebook, Twitter and LinkedIn.

Already, scouring Facebook and other social network pages of the insureds is a common practice on the claims side of the business. Many investigators say it's one of the first things they do when looking into potentially fraudulent claims, including both hard fraud (staging auto accidents, etc.) and soft fraud. (over-reporting damaged values after a fire, etc.)

Crack the Data

Currently, social network data are being used as sources of evidence in courts of law in claims cases. Individual underwriters are retrieving risk evaluation information on their insureds through manual searches on social sites.

But in a few years, automatically mined data from social networking sites could find their way into the underwriting pricing process. It could become a factor in determining premiums for both personal and business insurance, according to a new report from Boston-based research firm Celent. The report, titled "Using Social Data in Claims and Underwriting," was published on Oct. 10.

Could Offer Similar Insights as Credit Health

"Just as insurers recognize a link between credit health and risk in auto insurance, social data may offer similar insights for insurers who set out to crack the data," the report stated.

As users interact with multiple social networking sites, purchase items online, and communicate with others in public forums, they leave behind data about their preferences, lifestyle, operations and habits, according to the Celent report. This data can be used to develop a risk profile for an individual or for a company. On the corporate side, companies postings also include descriptions of new product offerings (hence new added risks), services and operations.

Connections and Links

Another piece of useful information is the "social graph," which shows how individuals or companies are linked together: a picture of who is friends with whom, who follows whom, and what friends of friends people have. In addition to identifying fraud organizations, these graphs can give insurers further insight into how an individual may perform as a risk, based on the behavior of those he or she is connected to.

Part of Underwriting in 3 Years

Use of social data is still in its formative stages, but it's developing rapidly. Celent predicts that over the next three years, social data will be "incorporated into core underwriting and claims processes" and become standard inputs into risk evaluation and settlement activities.

Celent contends that social data has the potential to join existing third party data sources such as CLUE (Comprehensive Loss Underwriting Exchange), motor vehicle reports, and MIB fraud reports to enable more accurate underwriting evaluation/pricing and lower claims costs.

Privacy Settings

The report says claims professionals have been learning how to link relationships on social sites to gain information on profiles which use the highest privacy settings. Investigators report that "People always have friends of friends willing to ‘accept' a new ‘friend,' and that friend's privacy settings aren't always set, allowing us to access pictures and the ‘private' claimant/insured's comments indirectly, detailing events they attend, groups they're associated with, etc."

While bypassing certain privacy settings is technically possible, it also raises a different question: what is ethically or legally allowed when mining social networking sites, and to what extent should they be used for insurance purposes?

Celent suggests one way to deal with this issue is engaging with customers, seeking permission to use their social data, and then automatically gathering it. Depending on regulatory rules, insurers could offer customers discounts as an incentive.

Challenges That Lie Ahead

The report said there are untapped analysis opportunities in tapping into recent advances in actuarial science, predictive modeling, and tools to analyze social data and to discover and leverage "hidden relationships in social data."