For decades, the insurance industry has relied on raw data to understand the financial impacts of risk and exposure, after all, that is the heart of their business. Being able to analyze and understand this information has always been fundamental to an insurer’s sustained viability; so to say data is an important aspect of what they do is an understatement.
Over the past few years, advances in computer technology have led to an explosion of new data sources for insurance companies to investigate. But is having access to all this data a good thing? Not necessarily, says Dun & Bradstreet’s Head of Global Analytical Services, Jayesh Srivastava. “You must understand the good data from the bad,” he explained while speaking to over 100 executives from various insurance companies at this year’s Predictive Modeling Conference in New York City. “It’s important to start with the purpose of the data and ask yourself ‘how can this data help solve a problem?’
In a world awash in data, it’s more important than ever to plan and execute initiatives with a thorough understanding of the quality of available data. After all, having enough information to make a decision is not necessarily enough information to make a good decision. Data has no real value if it does not help answer a specific question. For insurance companies it’s vital that the data is not only accurate, but that it can help confirm assumptions. Insurance patterns are notoriously unpredictable, with extreme variances based on everything from environment to population, so it is important to understand how stable the data is, as well as the source of the information. “Don’t take the data at face value,” said Srivastava. “You really need to understand where the data came from to gauge its value.”
With so much data in front of us, some of the most important sources of data for the insurance industry can range from export/import data to help gauge the viability of your supply partners, global growth data, which effects insurance pricing, and even global credit scores, to understand the health of a business. You should be able to link this data to meaningful observations, so it’s important to go after the sources of data you can link to the type of things you can do. Of course, it takes time and effort to transform this information into consumable insights.
Ultimately, data value is stunted until it is disseminated in a digestible format that provides context at a regular frequency. Today’s modern insurance organizations require custom-tailored reports, dashboards, and business intelligence (BI) platforms that provide interactive insight which helps to ensure mission success. It’s really about the insights that can be driven from the right data sets. Srivastava has a few tips for insurance leaders in their data-driven mission:
- The data you gather needs to connect to your other data sources to ensure it is accurate. It should always tie back to a reliable source to get to a single version of the truth
- Understand the stability of the data, it will impact how you analyze this information
- Data is not unbiased; you can’t believe data alone on statistical faith
- Dig below surface. Understand where the data is coming from and how it’s being serviced; trust the sources
- The data you consume has to meet your specific taste—ensure it is useable in your systems and can mix well with those sources
- There will always be missing gaps when you’re’ dealing with so much data; identify which segments the data impacts
- Think about the unstructured data you collect, like all the social data signals. Some of it can be useful, specially using it to understand consumer sentiment, but understand it is not perfect yet. Reducing noise is a challenge. Also, it’s not easy to scale…for now
- The data pool is vast and constantly evolving, consumption requires ad hoc processing
- While it’s important to have trend data and benchmarks, you need to leverage real-time data to understand where new patterns are emerging; your current models are evolving quickly
- Visualization helps get executive buy in, especially for trend data. It’s important to embed in all aspects of your analytics—it can help the larger team see the benefits and understand the complete story