In today's challenging global economy, insurers face thedifficult task of controlling costs and allocating resources andcapital effectively, all in the face of intense, risingcompetition. Generating additional business lift more efficientlyis a high priority, and insurers are increasingly relying on newdata sources and advanced analytics to do so.

Text mining is one such capability that property and casualty(P&C) insurers are increasingly using to uncover criticalinsights from unstructured data sources. Using text mining,carriers are able to identify high-impact opportunitiesselectively, control losses and costs, allocate resources, andoptimize financial outcomes. In the back office, text mining allowsthem to gain a better understanding of their claims operations,minimize missed financial opportunities, optimize workflow, detectpotentially fraudulent claims, and derive actionable insights fromcustomer feedback, among many other benefits.

Enhancing Claims Effectiveness
At the center of the claims-handling process, adjusters have anunenviably difficult task. They must not only accurately evaluatelosses and negotiate fair and equitable settlements but also assessclaims for a number of additional routing opportunities, such assubrogation, fraud, independent medical examinations (IME), andcase management. Claims adjusters' knowledge, expertise, andattention to detail are crucial because any missed opportunities orpoor decisions can significantly affect claims outcomes andbusiness results.

Time management and multitasking are prerequisites for today'sworker but the demands on an adjuster are arguably more pronounced.For most claims, the adjuster relies on the insured and/or theclaimant to provide information and documentation, which might comein small increments or all at once. In addition, the multitude ofclaims that adjusters manage simultaneously creates unique workflowpressures that can lead to incomplete facts and missed actions.

Following an investigation, for example, an adjuster mayconclude that the loss suffered by an insured was caused by ahit-and-run driver, and he or she may therefore intend to find outif the police have identified a suspect. Because of the workflowpressures mentioned earlier, however, that well-intentionedadjuster may forget to add a "diary" for this action, and thefollow-up may never actually happen. Similarly, even if theadjuster discovers that the police have identified the suspectdriver, the adjuster may miss subrogation on the claim. Finally, inthe event the claims adjuster determines that subrogation exists,he or she may very well miss sending the case to the recovery unitbecause of the workflow pressures.

Eliminating Oversights
Text mining can help alleviate such oversights by automaticallyanalyzing the adjuster's notes and sending alerts aboutopportunities for action. As a simple example, a text mining systemmay recognize the phrase "hit-and-run driver" and automatically adda diary entry for the adjuster to "please identify hit-and-rundriver." Another text mining engine might closely followhit-and-run cases and seek to detect the concept of "adverse driveridentified" in the ensuing adjuster notes. Once a detection hasbeen made, the claims adjuster could automatically be prompted to"please evaluate subrogation opportunity." Such text mining alertscan significantly improve the efficiency of the adjusting processwhile drastically reducing the number of missed opportunities foradjuster action.

Although setting up a text mining system canbe challenging given the unstructured flow, typos, andabbreviations in adjuster notes, the eventual business results arewell worth the effort. Furthermore, once a core text mining engineis established, it can be quickly and easily extended to produceinsights across a variety of claims concepts, from identifying theat-fault party to fraud red flags, dissatisfied claimants andinsureds, and details about the location and the cause of a givenloss.

Mining for Fraud Mitigation
Fraud is a challenge that pervades all industries, especiallyP&C insurance. Estimates from the Insurance InformationInstitute (I.I.I.) indicate that fraud accounts for approximately10 percent of the P&C insurance industry's incurred losses andloss adjustment expenses (LAE)—or about $30 billion each year.Recent indications also show that fraud is a growing menace inpersonal injury protection (PIP) states, with direct impact oncompany results and insured premiums. Recognizing, investigating,and denying suspicious claims are thus key priorities forcarriers.

A common example of automobile insurance fraud is the "swoop andsquat." This refers to a scheme in which several potentialfraudsters use an older car to cause an accident with a newerluxury vehicle, based on the assumption that the owner has highinsurance coverage limits. The scammers pull in front of theirintended target vehicle—which is usually near a highway exit—andstop suddenly, thereby causing a collision. They then report damageto the car and fake numerous soft-tissue injuries, leading to a bigpayday for them and a big loss to the victim's insurancecarrier.

While experienced, alert adjusters can recognize such schemesand get the Special Investigative Unit Services (SIU) involved in atimely manner, workflow pressures, lack of experience, andinadequate training can often lead to missed opportunities.Furthermore, fraud schemes are constantly evolving, and patternsmay not be as easy to detect as in the above example.

Text mining can be leveraged to distill insights from adjusternotes to systematically create a multitude of fraud concepts, suchas questionable injury, excessive treatment, low-speed accident,sudden stop, and an accident near a highway exit. Predictive modelscan then be established using these concepts—along with otherstructured data from the claims system—to generate highly accurateand actionable referrals for IME or SIU intervention.

Improving Workflow Routing
In addition to alleviating flaws in the claims process, text miningcan help reduce blemishes in workflow. Many insurers continue touse legacy claims-handling systems. An adjuster may identifysubrogation opportunity on a claim, but to route it to the recoveryunit, the adjuster may have to navigate through a number of screensor pages to the recovery referral screen. Worse yet, at somecarriers, the adjuster may have to open a separate database andcut-and-paste information into it to route the claim to therecovery unit. Owing to this obviously burdensome process, manyvalid referrals are often missed.

Text mining, coupled with strategic phrase training for claimsadjusters, can eliminate inconsistencies and improve the workflowprocess. Managers can train their adjusters to enter apredetermined set of phrases into their claims narrative, and anightly/batch text mining system can be set up to automaticallyidentify and route the claims for appropriate business action. Forinstance, the adjuster might write "*** Route to Recovery ***" inhis or her claim notes as the signal to the text mining system tosend that particular claim to the recovery unit.

Simple applications of text mining such asthis can increase adjuster productivity and potentially improve acompany's effectiveness in handling previously missed subrogationopportunities.

Manage the Customer. Innovate theProcess.
Customer contact areas at insurance companies typically deal withissues other than claims. The issues can range from billingconcerns to policy- and coverage-related inquiries.

Traditionally, these comments are buried in CRM-like systems andseldom leveraged. With increased use of e-mail, voice-to-textapplications, and social media sites, however, the volume ofcustomer commentary that can be made available to be mined isincreasing. The information derived from these contact points isrich and may be used to increase customer satisfaction andretention, improve business processes, and even identify areas forinnovation.

Customer comments can be systematically text-mined to discoverpersistent issues and new ideas using sentimentanalysis—essentially the ability to discover opinions and tones intext. For instance, customer complaints regarding the billingsystem may reveal a sentiment of dissatisfaction through words suchas "unable, down, terrible, and slow."

Armed with the type of issue and the sentiment, an insurancecarrier can take proper action. The complaint can be routed to theappropriate customer service representative to resolve thecustomer's issue quickly and create a positive customer experience.In the longer term, persistent patterns of specific issues can beused to make the business case for prioritizing resources andbudget investments. For instance, multiple customer complaintsmight help focus the carrier's attention on improving the billingsystem, while customer suggestions for an iPhone app to handleaccount management and claims might trigger new product developmentinvestments.

Critical Business Insights
In sum, text mining is a versatile analytic capability that offersnumerous benefits to the P&C back office, from effectivelyanalyzing losses and routing claims to combating fraud, minimizingmissed opportunities, and enhancing the customer-contactexperience. These techniques are quickly transforming obscuretextual data assets into insightful business-critical applicationsthat enhance claims efficiency, customer satisfaction, andresponsiveness—cornerstones of customer loyalty and retention.Insurers that successfully adopt text mining can significantlyimprove back-office functionality and effectively provide theircustomers and policyholders with excellent service and favorablereturns for the company.