By Phil Hatfield
If there’s one thing insurance companies have a lot of, it’s data. There’s quote data, sales data, policy administration data, billing data, reserving data, investment data, and claims data from internal systems. And outside sources can encompass government (census data, Federal Reserve data, Bureau of Labor Statistics data), weather patterns and history, credit data on both consumers and businesses, claims history, and market segments. The list seems endless.
Chances are your company spends millions of dollars a year and a lot of time and effort to collect this data and store it. And the rate of data accumulation has been increasing exponentially in recent years. The real issue then is: Does your company use all that data? Does it make you smarter? Does it help you reach decisions that positively affect your bottom line?
If your company is like most, your answer is no. So why, in an industry that collects so much data with ever-increasing speed, do so many people feel they lack adequate information? Because data is not the same as actionable information.
In its simplest form, a piece of data is merely a fact — a characteristic associated with a subject. For example, if a company associates policy A1234 with ZIP code 74101, it creates a piece of data that might be stored in a database as A1234/74101, where policy A1234 is the subject and ZIP code 74101 is the fact associated with Tulsa. Other pieces of data associated with policy A1234, such as address, name of insured, coverage type, and policy effective date, could be grouped together to create a policy “record.” If we had many policy records with the same type of data in the same format, we could combine them into a policy table. This is typical of the work IT does to arrange data into structures that are easier to catalog, manage, update, and retrieve.
But as mentioned before, data for the sake of having data is usually not very useful. For example, “It was sunny in Tulsa on Monday” is a piece of data that might be stored in a database (Tulsa/sunny). On its own, this fact may or may not interest you, and it probably won’t help you do your job. But if we combine this with other facts, such as “John Q. filed a claim for hail damage to his car on Monday” and “the garaging address on the policy is in Tulsa,” we have some information that will inform our actions about whether to settle the claim now or continue looking for more data. We should probably investigate further (specifically, look for more data). In doing so, we find the claim location was Little Rock, where he was visiting, and there were severe thunderstorms in the Little Rock area on Monday. We then decide we have enough information to settle the claim and pay John Q. We were able to combine some relevant facts into a logical scenario that gave us a good picture of what was happening. This is one type of what we typically call “analysis.”
We spend considerable resources and energy collecting and storing data just because the industry produces a lot of it and because the regulatory environment mandates storing much of it. The industry doesn’t spend nearly enough effort aggregating its data across functional silos, integrating internal data with third-party data, analyzing the data, and distributing the resulting insights to people who can take action on it. Using the simple claims example above, ask yourself: Would a claims adjuster at my company know the garaging address for the policy listed on the underwriting system? Let’s say maybe. Would he or she know what the weather was in Tulsa on Monday? That’s unlikely. Would all these pieces of data come together at the same time and be presented to the adjuster in a way that’s easy to analyze? That’s highly unlikely.
The root and original meaning of the word “analysis” is to break something down into its basic or constituent parts. That’s what IT usually does to aid in efficient storage and management of data. However, the difficult part — and the real art of creating actionable intelligence — is “synthesis,” or combining disparate pieces of data. Data that you can link, combine, and aggregate in new and novel ways can lead to original insights and informed decision making.
Nowadays when people speak of “analysis,” they typically mean the activity that turns data into insights. We can bow to convention and use this more common definition of analysis in our everyday speech. But keep in mind that IT has usually already thoroughly analyzed the data before entering it in a database. The analyst’s job is really to synthesize the individual data components into useful insights.
There are several steps insurers need to follow to achieve actionable insights from their data:
Unfortunately, following the steps isn’t easy or linear. For example, identifying a KPI in step 4 may lead to the discovery of a data gap that will require a return to step 1 to start again — at least for that data element. Also, it’s not a once-and-done task; it’s a continuing process. As the insurance industry and technology change, so do a company’s information needs. Senior management will need to make a major commitment because the scale and scope of the impact on the company is so large.
But if you believe better information leads to better decisions, then the effort should yield large rewards. The mere fact that the industry often uses its data inefficiently means insurers that discover how to optimize their data will have a significant opportunity — and if they keep up the good work, they’ll also keep their competitive edge.
Phil Hatfield, J.D., CPCU, is vice president of operations at ISO Innovative Analytics.