As insurance companies adopt complex, automated tools to improve their business processes, they must rely on individuals to use those tools correctly. Unfortunately, natural bias, human error, and common statistical mistakes often contribute to individuals making incorrect decisions. However, by embedding analytics into core operations through the use of technology and process design, insurance companies can train their workers to compensate for their biases and more consistently make better decisions.
A number of insurance companies have developed automated systems for managing various functions of the claims process. But the potential value of any analytic process still depends on each claims adjuster’s judgment and adoption of the available support tools. Critical decisions made by a claims adjuster are affected by the adjuster’s natural biases and the potential for human error. By operationalizing analytics, insurers can better equip decision-makers to apply quantitative thinking, along with cutting-edge technology, to their work, resulting in more informed decisions and better business outcomes. Because claims is the greatest cash outflow for insurances companies, even marginal improvement in the indemnity and expense of paying claims results in a large financial payoff.
How data is revamping insurance processes
While the use of Big Data and analytics across the industry has yet to reach maturity, the opportunity it provides is truly transformative. New technologies offer the ability to harness various types of live data that can be automatically captured, processed, analyzed, stored, shared, and used to assist complex decision-making in numerous business processes. Many insurers have already begun adopting advanced data analytics to improve various aspects of their claims functions. Predictive analytics is improving fraud programs by making Special Investigation Unit (SIU) case-tracker systems more accurate, easier to manage, and more effective at prioritizing a higher volume of cases while minimizing false positives. In the area of subrogation, insurance carriers can optimize collection and reduce the number of days and costs to collect by utilizing subrogation identification, segmentation, and collection- effectiveness models. For instance, a predictive model can identify subrogation potential, while other models can predict and maximize recovery amounts and collectability. Analytics is also being used to improve the First Notice of Loss (FNOL) process. For example, automated segmentation with embedded analytics enables less-complex claims and first-party coverage handling to be addressed as part of the FNOL process, reducing consumer and carrier effort during intake and improving customer satisfaction.
In the life and health insurance spaces, some insurance companies are already using external biomonitors such as fitness bands and wearable technology to track health and fitness data that could help improve customers’ health and reduce claims incidents. For instance, such technology tools allow insurance companies to collect information or send alerts when policyholders experience abnormal pulse rates, blood pressure, oxygenation, or temperature; inadequate levels of exercise; or poor sleep patterns. While tracking such data is only possible with policyholders’ permission, it can result in discounts for users and, in the future, aid in the prediction of certain health incidents. While insurers are beginning to invest in these areas, it isn’t always in a structured manner that takes into account the entire claims handling lifecycle. Even with the right tools and data, the critical component that is missing from most data analytics programs is organizational change that ingrains analytics into specific processes.
Power still rests with decision-makers
Although new technologies make it possible to harness customer data and improve various insurance business processes, the individuals who oversee and manage claims still have control over the outcome of each claim. Each adjuster or business unit manager brings his or her own business experiences and personal biases to every claim. Even when the claims process is somewhat automated, individual adjusters make decisions at various steps along the way, such as when initially assessing claims, negotiating liability, or evaluating fraud. And those personal decisions are often fraught with bias and susceptible to common statistical mistakes.
All individuals have natural cognitive biases that affect their decision-making processes. These biases are not necessarily negative; they are simply natural tendencies. But in the insurance business, these individual biases detract from a company’s ability to rely on standard claims decisions across the board. For instance, when claims adjusters or business unit managers are initially presented with a claim, the availability heuristic may influence how or when they respond to the claim. The availability heuristic is a person’s tendency to judge the frequency or likelihood of an event by the ease with which relevant instances come to mind. In some cases, it’s appropriate to pay and close a low-complexity or low-value claim quickly in an effort to avoid frequent overpayment and reopened claims. If an adjuster remembers closing a similar claim recently, he may be likely to simply close and pay the current claim. But not all claims that seem similar are truly alike. Rather than leaving the decision of whether to quickly pay and close a claim up to an adjuster’s whim, analytical data can help make such decisions more standard and reliable. For instance, using a decision support model, a claim can be assigned to a homogenous group based on its fine grain characteristics, ultimate losses can be predicted, and anomalous features can be detected and described.
Claims adjusters are also often influenced by the probabilities and risk bias, which refers to the natural bias towards overstating the probability of certain kinds of bad outcomes. For instance, when processing a claim, an adjuster must rely on his or her own judgment and experience to estimate the case value, or the amount for which the claim will ultimately be settled. By learning how to recognize and avoid his own probabilities and risk bias, he can rely on decision support tools to ensure that the initial reserve accurately reflects the ultimate settlement value of a claim and will not be revised upward.
Operationalize analytics to overcome bias and improve results
Regardless of the fact that personal bias and recent claims experience can greatly affect claims decisions, adjusters can learn to make decisions with more standardized, reliable methods. That journey begins with the reimagining of processes to better leverage technology and enable decision-makers to better embrace analytics. Insures must think of this process as occurring over a continuous loop that is designed to continuously improve their operations by more effectively steering data into insights, and then actions. For claims adjusters, the process could look something like this:
While some decisions made throughout the claims process are routine and based on rules and calculations, many of the decisions have no rule-based solutions. When adjusters encounter these issues, expert problem solving and complex communication skills are important, but mistakes will still be made. This is why redesigning the process to promote continuous improvement through the intelligent use of data is critical. When adjusters and other decision-makers understand how to make decisions more analytically, identifying and mitigating the influence of personal bias, they are then ready to employ data support tools. At that point, a wide variety of decision support models, visualizations, enriched data, and historical statistics can be utilized to streamline the claims process and ensure better business outcomes.
Focus areas for claims management
There are several key processes within the claims handling function where continuous improvement anddecision support training can have a significant impact. For initial claim assignment the availability heuristic can be mitigated through greater reliance on segmentation, loss prediction, and anomaly detection techniques.The result is that insurers pay and close a low-complexity/low-value claim under the appropriate guidelines while avoiding overpayments and reopened claims. Another area of focus is initial reserve setting, which is often subject to probabilities and risk bias. Here, once again, by creating processes that allow staff to more fully embrace data analytics, initial reserves will more accurately reflect the ultimate settlement value of a claim and are less likely to be revised upward.
In the areas of fraud determination and subrogation, claims personnel are often subject to confirmation bias—the unconscious reference to perspectives that confirm a pre-existing view. For both functions, this bias can often increase the number of false referrals, limiting the effectiveness of SIU and subrogation processing groups.
Other processes, including claims settlement and inventory management, also benefit from decision support models. In the case of claims settlement, there is often a gambler’s fallacy, which places too much emphasis on previous events by incorrectly assuming that they will influence future outcomes. However, by redesigning processes to better utilize applied game theory, crowdsourced decision-making, and anomaly detection, insurers can more effectively come to a settlement agreement with adverse parties that minimizes negotiable payments, avoids litigation, and satisfies the policy holder.
For inventory management, where the goal is to minimize expense, cycle time, and negotiable indemnity, while also improving customer satisfaction, the bias is the same. The solution lies in developing a decision support model that identifies the optimal claim priority in a multiconstraint model that incorporates variations in claim arrival and complexity.
Putting it all together
Insurers should build frameworks that will help claims adjusters and business unit managers overcome their inherent biases and realize the value that data can bring to their decision-making process. These frameworks should be extended to specialized training modules that will help those who make critical claim processing decisions employ complex cognitive thinking and analytical support tools to improve business outcomes.
These training modules will help insurance professionals to better understand at which points in the claims process they should employ expert decision-making skills, and how using those skills can result in a more reliable, standardized flow of claims. In addition to training front-line decision makers, a program should also offer skills training, which enables supervisors to better equip their teams for making decisions, and prepares them to utilize the latest analytics tools to improve their work. While this is just one example, the need for better alignment between core processes and new technologies is an issue across the industry. As insurance companies grapple to take advantage of the growing possibilities of data resources and analytics tools, it is vital that they are able to ensure that their front-line professionals have the capabilities to make reliable decisions and use tools more efficiently.
For more information, contact, firstname.lastname@example.org and visit, genpact.com/what-we-do/industries/insurance/claims-management