Improving Business Results with Text Mining

By Karthik Balakrishnan June 14, 2013

In the property/casualty insurance industry, uncovering valuable information in unstructured data — such as paper files, faxes, documents, adjuster notes, and underwriter comments — can be like trying to find that well-known needle in a haystack. New technologies for text mining, however, are making it possible for insurers to use seemingly unstructured data for business advantage.

Text mining can help carriers identify, understand, and measure the root causes behind losses to enable smarter underwriting. “Water damage” is a great example, because often a claim possesses that descriptor and has no further detail in a claims system. Using appropriate text mining on claim adjuster notes can reveal whether the claim was the result of a “burst pipe” or a “weather-related event.” Applying that granular insight to identify and quantify trends can lead to development of smarter underwriting criteria and establishment of stronger loss control and prevention mechanisms.

Text mining can also deliver immense value in claims operations. In addition to assessing, evaluating, and settling losses, claim adjusters also have to make numerous critical determinations, including subrogation potential, suspicion, need for an independent medical exam, need for a professional engineer on a property claim, and so on. Making those assessments correctly can have a significant effect on business financials by helping insurers avoid missed opportunities and deploy costs efficiently. For instance, using text mining to uncover red flags such as “overtreatment” or “unnecessary treatment” can aid in assessment of suspicious claims and expedient routing to the special investigations unit (SIU). Unfortunately, according to a recent study by the Coalition Against Insurance Fraud, only 40 percent of insurers are using text mining to help prevent fraud. That points to a large gap in missed opportunity and value for property/casualty carriers.

Text mining has become a practical and highly useful capability to reveal critical insights from any kind of textual content — well beyond adjuster and underwriter notes. Given the plethora of textual information in insurance — including customer service call notes, loss control reports, and premium auditor reports, to name just a few — carriers would do well to contemplate innovative applications that can create important business gains.


Karthik Balakrishnan

Dr. Karthik Balakrishnan has more than nine years of predictive modeling experience in property/casualty insurance, building and directing the development of solutions for marketing (customer segmentation, channel/producer segmentation, opportunity scoring), underwriting/pricing (personal auto, homeowners), and operations (claim fraud detection, subrogation identification, premium audit prediction).