Insurance Analytics - Barriers to Exploitation

By Douglas Duncan, CIO, Columbia Insurance Group

Douglas Duncan, CIO, Columbia Insurance Group

Insurance analytics, and in fact all business analytics, is about turning information into money. Monetization. How effective is this, and what are the barriers?

Most companies are actively trying to improve this capability. According to a Novarica study, “Insurer IT Budgets and Projects 2018”, for P&C Carriers, nearly 50 percent of mid-size and over 60 percent of large carriers surveyed are planning or implementing major enhancements and replacements in this area.

"When we change our environments we impact our ability to extract meaning from the data"

Companies are making significant investments to improve this capability. Assuming the investment pays off and better information is available, how do we monetize this? Generally speaking, it is through superior pricing of risk, better understanding of a market or product, improved detection of fraud and waste, and increased efficiency of operations. Master these four topics and you can rule the world. In practice we face many challenges.

If we can understand our information better, we can make better decisions based upon it. This is conventional wisdom, and is a great starting point. Why is this not happening as effectively as we wish?

The Standards Proliferation

Have you ever tried switching from one type of data warehouse to another? From one reporting and analytics suite to another? It can be messy, painful, and expensive.

When we change our environments, we impact our ability to extract meaning from the data. Every tool builds a framework around the data in order to properly analyze it, which requires extensive effort. Changing tools and environments is costly and often we lose nuances that are important.

Artificial Intelligence and the Data Lake approach to storing and parsing huge sets of unstructured or poorly structured data promise a solution, but the technology is not yet mature.

The Information Torrent

A common complaint in the past has been in not having enough information to make a good decision. These days it is more accurate to say we struggle to separate out the wheat from the chaff. We need to determine which information is relevant and which can be safely ignored.

Applying a data sciences approach to understand our information better is a hot topic, as evidenced by increased hiring activity. Not every company can or will bring on a data scientist, though someday it may become very common. However, short of hiring for this role, every company can still leverage best practices in data analytics and utilize third party support to improve how they manage their data and harvest the value.

The Reporting Conundrum

You build a shiny new data warehouse. You implement a slick new reporting and analytics tool. You have a team in place to put this all together. Now you can sit back and enjoy the flow of meaningful, actionable information…in your dreams.

You need to consider three reasons why your new data warehouse and tools may not provide the insights you need. First, you may not have a clear concept of reporting versus analytics, which muddles how you build and optimize. Second, you might focus too much on re-creating the reports you had previously sourced elsewhere, resulting in the same information without new insights. Third, you perhaps treat reporting as a static thing, and under-utilize the concept of dashboards and reactive reports.

Reporting and analytics should not necessarily be separate disciplines within your organization as they are both about data exploitation. However, do consider them as serving different purposes, utilizing different tools and organization of the underlying data. Good analytics is dependent on NOT treating it as only an extension of reporting.

The Ownership Paradox

Who owns your data? An older and more technical answer is IT. Most people today would answer that it is the business (the European Union will say the customer). This is an important question. Who takes better care of a house or car? Clearly the owner has a vested stake in the product and a renter only cares about the current utility. Is the same true for data?

Here is the paradox. The care and grooming of data is today a highly technical process. It requires specialized tools and knowledge. The expert has to stay close to the data to understand the needs and to make it as exploitable as possible. However, the use and understanding of data is also a highly specialized process, requiring sophisticated skills and experience. Both skill sets are critical to data exploitation success, yet they often exist in different parts of the insurance business, most often IT and Actuary.

As with the Information Torrent, Data Sciences might be able to ride to the rescue. This discipline can bring together use vs. maintenance, renting vs. owning and building vs. exploiting. If we do not ultimately combine these aspects, the conflicts created by mixed ownership will continue.

Monetizing Information

Monetizing information has always been the goal of insurance analytics, but it is becoming the defining principle of services industries in general. Thinking about and addressing the Standards Proliferation, the Information Torrent, the Reporting Conundrum and the Ownership Paradox will help you improve your approach to insurance analytics. Adopting a data science approach, else adopting best practices where possible, will be a big step in the right direction.

Weekly Brief

Top 5 Insurtech Solution Companies in UK - 2020
Top 10 Insurtech Companies in Europe - 2020

Read Also

InsurTech and the Benefit to the Business

InsurTech and the Benefit to the Business

Brendan Mills, CIO, nib Group
Artificial Intelligence and Business Productivity

Artificial Intelligence and Business Productivity

Greg Tacchetti, CIO, State Auto
Digital Strategy

Digital Strategy

Douglas Duncan, CIO, Columbia Insurance Group
A Promising Road Ahead for InsurTech

A Promising Road Ahead for InsurTech

Chris Purcell, CIO, PEMCO
Seeking Analytic Diamonds In The Underwriting Data Mine

Seeking Analytic Diamonds In The Underwriting Data Mine

George Di Martino, EVP & CIO, TransRe
Transformations in Financial Technologies

Transformations in Financial Technologies

Jacob Sorensen, CIO, Bank of the West