Swiss Army knives are designed to do just about anything--from cutting to sawing to drilling to poking to scraping and on and on. But when something new needs to be done, if the gadget isn't there, you can't modify the knife to add new tools. Like a Swiss Army knife, many companies intend for their insurance policy administration systems to do just about anything. Software vendors and systems designers make their living ensuring nearly every feature an insurer might need is included in their product or can be custom added.
Although this paradigm may provide insurers with power and capability, experience warns us to be mindful of systems that become overly bloated with the 20 percent of things contemplated in the 80:20 rule--while some of the things desired for the 80 percent bucket may not be in the system. This is especially significant when new major business or product capabilities evolve, and it becomes essential for them to become rapidly incorporated in policy management systems--capabilities that are inherently part of the 80 percent of needs, capabilities such as predictive analytics. But like a Swiss Army knife, what do we do if our systems can't be readily modified for new required functionality such as analytics?
A simple trip to a bookstore informs us of the mainstream importance of predictive analytics as we observe a startling number of topical pop-culture books on display. In our industry, much has been written in the insurance literature about predictive analytics and models that help insurance executives better analyze and predict circumstance and outcomes--things such as the potential profitability of new and renewal customers and risks. For some companies, these models have become part of the essential fabric of CRM, underwriting, pricing, and agency/broker management, to name a few. There is no doubt predictive analytics has become a mandatory capability in today's insurance world.
New analytic solutions continue to be developed, and existing ones continue to be enhanced and revised. However, few policy management systems have evolved to easily make the connections to these analytic tools more streamlined and to allow for rapid development and update of predictive tools when they are integrated with policy systems. Typically, when changes need to be made, it's a complex, time-consuming, and expensive effort. ROIs are analyzed, critical paths are reviewed, and priorities are juggled. But if policy systems could more easily adapt to necessary changes in predictive models and other analytic components, the significance of policy system maintenance would be greatly diminished. With the ever-growing application and enterprisewide integration of predictive analytics within insurance company operations, the need for fluid and dynamic policy administration systems becomes even more essential.
Without a comprehensive systems and data architecture for predictive analytics' role in a company's business and operations, companies will continue to manage such solutions as "bolt-ons" to core policy administration and ancillary systems. If this continues, an unfortunate byproduct could be significant missed business opportunities because predictive solutions are rapidly evolving to become part of the DNA of many insurance products and processes, and companies risk not being able to keep pace with their competitors that are making progress in this area. Market leaders today will be defined as those organizations that can process and distribute information internally and externally faster than their competition. Furthermore, rapid product development is a necessity, not a "nice to have," with mass customization of traditional products and their features along with hybrid products becoming more commonplace.
Companies are under enormous pressure to be highly responsive to new and existing customers and to their distribution partners. Predictive analytics solutions have evolved as important components to the business and to accomplishing these goals. Many companies have stated objectives to develop a more data-driven culture and to better develop the science of insurance in addition to the traditional art of the business. As companies work to execute on these objectives and modernize their approach to bring science more to the forefront, it becomes essential to integrate analytics flexibly with policy management processes through the development of reusable utility services that can serve a broad array of applications.
Two examples of such analytic services are scoring engines and rules engines--both of which embody the intelligence of predictive analytics and then translate the signals into succinct business information and actions. These scoring and rules engines need to accept data from policy management systems and return output data back to them. The handshakes that need to take place between these various services must be designed in a way that allows for easy addition and revision of functionality, as models and rules change frequently within the engines. Companies must move away from designs that require resource-intensive projects and large-scale systems modifications just to update an analytic model, its data inputs/outputs, or the actions resulting from the execution of the rules from the model's signals.
Service-oriented architecture (SOA) provides many of the answers to the need for easier and speedier process change required in today's business and systems climate. SOA allows for the development of distinct and separate systems functions to be joined together seamlessly, like building blocks, into an integrated solution. SOA allows for essential actions, such as data acquisition, data hygiene, model scoring, model output, rules processing, and management reporting, all to be linked rapidly and efficiently through a common processing language and messaging protocol.
As a result, when a company develops a new product, tweaks an existing one, or must gather new customer information for some new analytic model, the organization doesn't need to spend enormous energies getting the job done. Through SOA design, gone are the days of changing "spaghetti code" and hard-coded logic and rules because well-designed SOA systems allow for many such changes to be made quickly. Functional systems components, such as product configurators, allow a company to assemble systems product functions like car companies assemble vehicles--based upon the common chassis, different functional parts and pieces are assembled to make analytic solutions rapidly come alive or be maintained more efficiently. Not only does this help with the introduction of new products and features, but it also helps prevent system stagnation as the cost to maintain the system is greatly reduced--maintenance and updates are more financially palatable.
As companies continue to innovate, it becomes increasingly important for their analytic models to change with the times. New data sources constantly become available to provide insights into risk. New processes make data available at different times in the policy and product life cycle. New regulations impose constraints or requirements on how business is processed and documented. And central to all this are models that help to drive better operations and decision-making--models used for multivariate underwriting, risk aggregation analysis, or portfolio management; models to determine the need to purchase expensive external data reports (MVR, CLUE, parameds, etc.); models to predict suspected application vs. exposure disparities; models evaluating the lifetime value of a client, etc.
In order to better appreciate the relationship of advanced analytics to policy management, let's quickly walk through a P&C underwriting example:
A company writes commercial workers' compensation (WC) coverage and has researched a variety of innovative pieces of information as potential predictors of the quality of new or current policies. The company's analytics team has created some new predictors for its multivariate underwriting model. One new piece of information it will gather going forward for use in its model is whether or not an employer company offers health insurance to its employees and the types of coverage offered. Underwriting management wants this information entered on the policy management system's underwriting screen because it has been shown employees without company health coverage disproportionately report small-case WC claims (perhaps because they may use the WC coverage as a substitute for health insurance). In order to capture and use this new information and accommodate the new data field in the policy management system, some next steps may include:
o Add the new data field to the policy system's database.
o Modify the underwriting screen and through the agent portal.
o Modify the multivariate underwriting model in the system scoring engine.
o Modify the data feeds to and from the scoring engine to include the new data field.
o Modify or add business rules to the rules engine to utilize the model results.
o Modify policy system screens to show any new information or results.
o Modify management reports or BI processes.
There are other tasks, as well, tasks such as modifications to automated data feeds to/from third parties, interfaces to agency management systems, linkages to data warehouses--these additional tasks make the systems changes even more complex. It is easy to appreciate how much work needs to be done just to insert a new piece of data into the end-to-end underwriting process, something that can significantly disrupt existing business processes if not planned carefully.
This example helps to show why policy management systems need to be designed to be more flexible to accommodate the needs of advanced analytics and to better serve the increasingly dynamic marketplace. Whether the system serves to manage P&C or life insurance policies, the broad concepts and functionality are the same. Each time changes need to be made to accommodate a new product, feature, or option and the related underlying analytic models, a company can't approach the event with dread knowing execution of the plan will require a massive, expensive, lengthy, multidisciplinary business and systems project.
Systems that are architected to allow for ready change and adjustment can generate a high ROI and actively enhance a company's profitability. Their ability to generate speed to market will allow carriers to compete more effectively with a higher level of constituent service. With tools to perform rapid prototyping and production modification, companies can better serve the most significant business needs, such as predictive analytics.
While many carriers' policy management systems may, metaphorically, continue to look like Swiss Army knives, there is a strong need for them to be more capable of adapting to the needs of the business and the changes required by the various analytic tools that are becoming so important for company success. Advanced analytics isn't just a passing fad, it's the wave of the future. Company survival depends on its systems being able to keep up with the times but without the need for constant heroics.
