Fraudulent insurance claims — the hits just keep on coming, withmore than 60% of insurers experiencing increases in such claimsbetween 2013 and 2016, according to researchreleased last year by the Coalition Against Insurance Fraud. InSeptember 2017, the InsuranceInformation Institute pegged the cost of fraudulent propertyand casualty insurance claims at $34 billion annually.

In addition to human resources to combat the problem, insurersare taking a different tack by adding software robots (also knownas robot-processed automation, or RPA, solutions) to their fraudinvestigation workforce. These solutions automate the process ofsifting through personal and other data found online tosubstantiate or disprove the fraudulent nature of any claim. Theseare just a few of the benefits of RPA solutions:

• Facilitating more thorough investigations ofpotentially fraudulent activities, at a lower cost. Unlikehuman investigators, software robots are relentless by nature. Theycan rapidly troll the internet for data that may point tofraud—including the ever-increasing volume of personal data nowbeing shared on social media outlets such as Facebook and oncollaboration platforms like Slack. The time and effort required tohandle these tasks, as well as the cost, pale in comparison to theprice of employing human investigators to tackle them. Even moreimportantly, software robots' use of algorithms in mining data forcorrelations allows them to identify patterns and other indicatorsof fraud not typically discernible by human investigators. RPAtechnology becomes increasingly better at drawing correlations withend-user training. When machine learning is incorporated withsoftware robots, the sheer volume of data collected and theintricacy of patterns identified is overwhelming—in a goodway.

Related: Capturing insurance's most wanted fraudsters

• Freeing up valuable resources to handle more complexcases. Software robots can uncover most of the so-called"grunt work" pertaining to straightforward claims—i.e., searchesfor policyholders' social media data and other information aboutpotentially suspicious activity that may appear online. We see RPAusage moving in this direction, as it frees up investigators tofocus on higher-profile cases and to follow up on or substantiateany instincts they may have about them.

Related: How AI and automation can transform the claimsprocess

• Reducing the overall incidence of fraudulentclaims. Certain individuals do not see the harm in "plainvanilla" insurance fraud, believing it is a victimless crime.However, policyholders pay the price of fraudulent claims activityin the form of increased premiums. Other incidences of insurancefraud are linked to financial crimes and terrorist activities. RPAmakes the activities of perpetrators in both camps easier topinpoint and more difficult to conceal, potentially dissuading themfrom pursuing false insurance claims altogether.

Related: How insurers can identify fraudulent claims before they arepaid

Best practices: Bigger bang for the RPA buck

Software robots clearly rank as a valuable tool that insurerscan leverage when investigating and helping to mitigate fraud.However, applying best practices in their use is imperative tomaximize investment in the technology. Read some of our bestpractice suggestions below:

• Don't expect "out-of-the-box" functionality.RPA is very much an across-the-board technology. It has a multitudeof applications within and beyond the property and casualty sectorsof the insurance industry. In order to capitalize on thefunctionality of any RPA solution, that solution must be tailoredto handle individual organizations' particular needs—and with time,it will become even better-trained to do its assigned job or jobs.A reputable vendor will offer expert assistance from its datascientists and machine learning engineers to ensure such anoutcome.

Related: 3 ways machine learning is transforming insurance

• Start with one small project. Implementing RPAacross the board in one fell swoop will prove too overwhelming forall concerned. Instead, find a smaller project—for instance, onethat involves a particular kind of investigation and doesn't crossbusiness lines. Collect the necessary data and assess whether it isadequate for its purpose. If not, determine what information shouldbe added and identify any other tweaks that should be made to thesystem.
• Keep legal defensibility in mind.Investigations of insurance fraud frequently lead to litigation.However, unless it can be proven that data uncovered using RPA hasbeen collected, archived, and retrieved in a tamper-proof manner,it will not be considered admissible evidence in a court of law.This, of course, defeats the purpose of investing time and money inRPA in the first place. For best results, choose an archivingsolution whose configuration removes the possibility of tamperingfrom the equation. Limit vendor selection to those that are willingto attest to how their archiving platform supports the legaldefensibility of any data stored within it.

Related: Using smart technology to combat insurance fraud

• Debunk the myth. As is true of many technologyimplementations, there will be concern among some employees thatthe launch of an RPA solution will cost them their jobs. Such fearis not unreasonable, but it is largely unfounded. While a fewineffective staff members may no longer be needed once thetechnology is in place, the vast majority of employees willcontinue to play an important role in the organization. The onlydifference: employees' time and talent will be harnessed moreefficiently, at a lower cost to the organization.

RPA is here to stay. Within the next 18 months, it will be thedefault business model for fraud investigation in the insurancearena. Organizations that fail to embrace the technology, even to alimited degree with the flexibility to branch out later, are likelyto find themselves at a great disadvantage moving forward.

Related: How RPA helps insurers remain competitive

Kevin Gibson, CEO, Hanzoan engineering company that provides "defensible collection of weband social media content."