In recent years, there has been much discussion about the merits of predictive analytics (PA) in project management that enables managers or business leaders to achieve scope, time and cost balance.

So, what is PA and how far it is used in the insurance world?

• Is it informative dashboard - MIS?
• Is it business intelligence?
• Is it an audit to find and reduce errors of a given and approved process?
• Is it parallel computing?
• Is it a new version of game theory?
• Is it a new way to achieve Six Sigma--faster, cheaper, and better?
• Is it quality assurance in a new avatar?

Our experience is that predictive analytics is a summary of all, particularly so in the insurance project management world.

We all know the five stages of project management--initiation, planning and design, execution, monitoring and controlling, and closing. While project management skills are obviously important for project managers, interestingly the methods and tools that project manager's use can be helpful for everyone.

A 'task' does not necessarily have to be called a 'project' in order for project management methods to be useful in its planning and implementation. Even the smallest task can benefit from the use of a well-chosen project management technique or tool, especially in the planning stage.

Project management methods and tools can therefore be useful far more widely than people assume. One such interesting tool is predictive analysis, especially considering its successful deployment in the insurance world.

According to Wikipedia, "Predictive analytics is a broad term describing a variety of statistical and analytical techniques used to develop models that predict future events or behavior. Predictive analytics encompasses a variety of techniques from statistics, data mining and game theory that analyze current and historical facts to make predictions about future events".

In the PA world of insurance project management, predictive models exploit patterns found in historical, empirical, and transactional data to identify risks and opportunities. The combination of data and modeling expertise gives the unique ability to create most effective predictive models for insurance industry. Better analytics and tools that incorporate new data points will continue to drive both lift and efficiency, translating into more proficient policy administration, improved cost estimation, smarter portfolio management and of course better service delivery. All this will lead to less claim ratio and operating profits for the carriers.

"In the earliest stages of adoption, predictive analytics was predominantly used in claims within the commercial property and casualty space, says Jim Haley, CMO of Valen Technologies. "Today, it is quite common to see insurers use predictive analytics to support claims reserving issues as well as to identify fraudulent activities within active claims".

Advanced analytics solution quickly identifies the root causes of errors in back office processes like intake, policy binding, underwriting, claims entry, claims adjudication, statement of health, customer care, etc. A mature PA tool could deliver 25 percent or more quality improvement in two to three weeks. Thus, leading insurance carriers and outsourcing providers are embracing the solution.

Why and how?

PA can reveal unknown underlying patterns to known quality problems that can help carriers, BPO firms or project managers quickly address these problems.

PA analysis is significantly more accurate than existing manual methods in detecting errors in the sample analyzed. Manual QC may not detect approximately 25 percent of the errors detected by PA

For the dental claim project, PA can go from raw data to finished analysis reports in less than 24 hours. A similar manual analysis would take significantly more time

PA not only reduces operating costs while further improving quality, it could open up a completely new value proposition

PA could enable a differentiated sales strategy or even become a significant additional source of revenues

Another big challenge is to convert the theory of PA into effective and sophisticated software algorithm. Unless we achieve that, all the virtues of PA will be on paper or at a limited space. Naturally, the moment we talk about software or IP platform, we are talking about set of bright and smart people who could convert the insurance business processes into software.

This IP should have flexible rules engine and could be placed on the given policy administration system, underwriting, or claims platform being used by the carriers. So, creating a state-of-the-art system is not good enough in isolation but it should be interfaced with the legacy system platforms or legacy insurance business processes.

It is interesting to note that PA systems evolve with time and lines of business they have been exposed to. We are not sure if one size could fit all. To dwell further, means that a PA system designed for health insurance may not work for life or property/casualty insurance. On theory it should, because we have learned the definition of "system" is 'garbage in and garbage out', so any PA system should work perfectly fine for any business or for that matter any industry as long as there is a historical or empirical data for the work flow.

Not really, though, because a PA system has virtues of artificial intelligence, thus it grows with time.

A shift is taking place. As insurance companies continue to become less product-focused and more customer-focused to gain competitive advantage, PA and advanced PA are becoming handy. It is easy to talk about competitive advantage or aspire for larger revenue vis-?-vis last year but we all know, it is not so easy. Perhaps the only solution is exceptional service by exceptional people. PA is needed to achieve that critical differentiation in the market place.

Practical Usage of Predictive Analytics in the Insurance World

Marketing

• Helps in direct marketing
• Helps in CRM
• Helps in cross-sell
• Helps in product prediction

Predictive Analytics (PA) software can comb through current and past customer service calls in search of speech patterns that indicate dissatisfaction. Carriers can home in on that data to determine what is driving customers to call. Policies and procedures can then be put in place to correct issues and increase customer satisfaction

Underwriting

PA helps combine clients' data with historical risk and econometric information in proprietary analytical database tools to build, deploy and monitor models for improving underwriting and optimizing premium audit processes

Sales and Distribution

• Producer compensation
• Training and guidance

PA helps engage more closely with their distribution outlets and partners to align and share new, sophisticated information

Pricing

• Pricing strategies and profitability
• Improve pricing strategies, risk selection and profitability
• Predictive analytics drives profitability gap

Claims

• Fraud Detection
• Helps in adjusting loss reserve funds
• Helps in prioritizing claims
• Identify high-severity workers' comp claims before they become costly, reducing settlement lags and claims payout
• Automatically assigning adjusters according to priority and skill set
• Helps in subrogation
• Helps in e-discovery and litigation support

Collection Analytics

• Delinquent customers who do not make their payments on time
• To ensure premium to exposure accuracy

PA provides the ability to predict the future outcome by analyzing the past pattern or data

PA is a great tool to predict correct pricing and highlight risk. To achieve true success of PA in the insurance space, actuaries, analysts and underwriters need huge computing power to run the sophisticated models

PA could be used by insurance leaders to respond effectively to natural disasters by repositioning assets and people.

Manish Jaiswal, ([email protected]) is national sales manager for insurance outsourcing and BPO solutions for Affiliated Computer Services, Inc., a Xerox Company.

Monika Vashishtha, PMP ([email protected]) is process and performance manager, Global Change Management Group, Thomson Reuters.