Are robots in your company's future?

Predictive analytics, for the sake of this discussion, refers to the collection, sortation, compilation and presentation of data in a way that makes it plainly understandable and useful in predicting specific insurance claims scenarios.

Unlike traditional actuarial analysis, which relies heavily on assumptions and usually lags in relevance, predictive analytics provides a more robust, scientific model of past and present information for insight into real-time claims processing decisions.

To the adjuster challenged with the task of quickly resolving insurance claims to a customer's satisfaction – while also ensuring the lowest payout at maximum recovery – predictive analytics can be a veritable game-changer. Consider the applications:

• Fraud identification: Mining claims for telltale signs of possible fraudulent activities.

• Litigation management: Flagging for swift resolution cases displaying the potential for escalated legal costs.

• Accurate reserving: Avoiding cost inflation and step reserving on long-tail cases based on accurate payout estimations.

• Subrogation opportunities: Spotting cases where liable parties might subrogate a claim to help recover claim expenses.

• Customer experience: Efficiently managing data to provide agents with real-time access to pertinent information to expedite resolutions.

Using computers to sort and identify pre-determined data sets can help insurers flag fraudulent claims. (Photo: Shutterstock)

A better use of time and resources

Putting talented claims adjusters on the task of compiling information is not an optimal use of time; machines work faster and yield greater results at significant savings. To envision the impact in terms of cost-benefit and user satisfaction, consider the following examples.

Fraud

According to the National Insurance Crime Bureau (NICB), insurance fraud ranks second to tax evasion as the costliest white-collar crime. The FBI estimates that more than $40 billion is lost to insurance fraud annually, not counting health insurance fraud, costing U.S. families upwards of $700 per year in increased premiums.

Fraud comes in many forms through many avenues, including individuals, businesses and crime rings, involving professionals such as doctors, lawyers, pharmacists, supervisors and other credible employees. The level of sophistication among fraudsters is high. As quickly as investigators unravel one scam, new ones hatch, creating a dangerous dynamic that costs carriers and customers alike.

Armed with accurate data at the right time, carriers can identify potential fraud through clever referral models that prioritize relevant cases for investigation. Uncovering complex patterns of known or suspected fraud, using historical records, social media, overlooked details and other background information can help identify suspicious individuals or groups that present previously identified negative behavioral traits or patterns.

Predictive analytics can help insurers recognize patterns based on predetermined rules. (Photo: Shutterstock)

Triage

At the operational level, the ability to effectively triage incoming cases for referral to the right adjuster not only streamlines workflow, optimizes assignments and expedites resolution, it can significantly lower costs by avoiding overpayment and possible litigation. Analytics models can be created to look at the nature of a claim, past data, organization skill set, injury type, severity and other factors to identify the appropriate resource to manage the claim.

Highly skilled resources, such as nurses or engineers, can be reserved for special cases, taking into account current bottlenecks and availabilities that ensure the best resource is on any given case. This can result in more intelligent and economical treatments in the case of healthcare claims, or less expensive materials and supplies in property damage claims, significantly reducing payout. It is typical to see annual claims payouts drop as much as 15% annually where predictive analytics is employed.

Assigning low-level clerical work to automated workflows and robots will allow talented adjusters to apply their experience, training and skills to solve critical claims and risk issues at a higher level of effectiveness.

Putting advanced automation to work

The road to creating an operationally valid predictive analysis capability involves customizing machine learning technology around a rules-based business model.

Machine learning describes the ability of a computer to recognize patterns and apply computational theories to simulate human-like intelligence. Customizing a rules-based machine learning business model relevant to claims and risk management decisions is essential to utilizing predictive analytics.

As you go about configuring a solution for your organization, examine the full claims lifecycle to identify areas where machine learning or robotic intervention could reduce manual labor or errors, uncover overlooked opportunities, create efficiency, enhance agent effectiveness or improve customer satisfaction.

Companies that do not have the in-house resources to implement a tailored predictive analytics solution can utilize a partner to compress the time, cost and disruption of building their own solution. The partner should have a proven implementation methodology and a track record of working with similar types of businesses.

The benefits of predictive analytics

Given the combined state of today's digital business environment and modern analytics technology, there is no reason for talented claims personnel to perform tasks better suited for machines. Let robots perform robot duties, while you empower your people to improve expense ratios, loss ratios, operating ratios and underwriting profitability.

Sean Allen (sean.allen2@exlservice.com) is vice president of EXL and a leader with over 18 years of business development and strategy experience in the BPO and ITO industries.