Data Analytics Is Changing The Insurance Industry

Several years of accelerating investment in data and data analytics are transforming the insurance industry. Data analysis has always been one of the historical pillars of insurance. Actuaries have used mathematical models to predict property loss and damage for centuries.

When they sell policies, insurers collect large data-sets about their customers that are updated when those customers make a claim. In recent years, as insurers have sought to become more relevant to their customers and more efficient, they have realised the strategic importance of their data investments. They want to harness data analytics to improve customer experience significantly, whilst cutting claims handling time and costs, and eliminating fraud.

Leveraging advanced analytics, and then integrating those results into their business processes needs to be an integral part of every insurer’s strategy.

Manually spotting troublesome claims early is challenging; working out strategies to mitigate the risk once identified is tougher still. The information needs to be delivered in a timely fashion, preferably instantaneously, into the natural workflow of the adjuster, possibly with a notification to the supervisor or large loss unit. 

The information delivered needs to not just raise an alert, but to explain the attributes which support the risk level, and propose a solution or work plan for the adjuster. This process should repeat itself in real-time as underlying data changes are made to the claim file, particularly for long-tail lines such as bodily injury.

Insurers are also turning to external data sources and adding more information about a claimant or injured party, such as identity verification or social media data. However, there are limits and barriers to just adding external data points.
Putting machine learning into how data is collected and analysed will help considerably in how insurers become more data-led and driven businesses.

There are some great examples of how insurers are using smarter predictive analytics to fast-track claims and process them with little to no human intervention. 

• A company called We Predict uses predictive analytics to enable vehicle manufacturers and suppliers to manage the frequency and cost of malfunctions for vehicles under warranty. 

• US insurer Esurance has taken to using predictive analytics as a means to skip adjuster inspections on motor claims related to major extreme weather events like 2017’s Hurricane Harvey.

The other data analytics issue confronting insurers is that actuarial science has limits when used to predict new categories of 21st century risks like cyber, food safety or complex supply chain disruption.  Helping businesses and individuals manage these risks offers huge potential rewards for the industry. Already global cyber premiums are growing at 30% per year with less than 15% penetration in the US and less 1% penetration worldwide.

Until recently insurers have not had the tools necessary to understand or price these risks accurately. There has just not been the data and loss experience for conventional modelling of these emerging categories of risk. While there are hundred-plus years of data about extreme weather, this is simply not the case for cyber-attacks, for example.

These new insurable risks have very different patterns and connections from risks to vehicles and property. For example, cyber risks are conjoined not by being in the same building or on the same flood plain, but by patterns of software usage, network connectivity, and human error

Even for well-understood risks, old assumptions may no longer apply. Indeed, the future may not be like the past. For example, changes in technology (semi-autonomous vehicles) and human behaviour (distracted driving) have already affected losses and their resulting claims in the familiar, well-studied area of personal car insurance.

How insurers will work with data in new ways is by embracing new models of technology, like Internet-scale “data listening” that aggregates, cleanses, and updates petabytes of data to build risk models in, for example, cyber. 

The data that is co-related might span public and proprietary sources about network presence, connected devices and control systems, and semantic content such as job boards and news reports of ISP outages and data breaches.
Ultimately, the industry is moving towards applying machine learning, natural language processing, and other modelling techniques to their core and third-party data in support of both operational and risk analytics. 

This ranges from underwriting tools for evaluating and pricing risk through the scores used to making better operational decisions in service and claims after those risks become insured.

Information-Age:       Image: Nick Youngson 

You Might Also Read: 

Will AI Make Data Analytics Jobs Obsolete?:

Cyber Insurance Risks Are Moving Too Fast For Underwriters: