Insurance companies talk about new product development and speed to market, but true innovation is rare in the insurance sector. However, connected, communicating devices, often referred to as Internet of Things (IoT) could be one technology to revolutionize the industry.
Sensors offer unprecedented access to granular data that can be transformed into assessing risk more accurately. For many insurers their initial exposure to IoT has been via telematics devices. But today sensors are used in thousands of different devices. Sensors are used in buildings and bridges to monitor for structural defects and mitigate potential losses. Life insurance companies are using the data from wearable devices like FitBit and Nike+ FuelBand to better assess the health of the life insured. And, sensors are being implanted into animals to track and identify livestock, helping insurers rate and price agricultural insurance more accurately.
The first sensors appeared decades ago, but in the last five years two major changes have shaken the sensor world and caused the IoT market to mature. From a technology perspective, the size and cost of the devices have decreased dramatically, and Wi-Fi and wireless communications make it more efficient to transmit the data.
In an industry that’s frequently slow to adopt cutting-edge technologies, IoT is starting to make waves. To successfully leverage IoT, insurers need to invest heavily in both data management and data analytics.
Big data has become a technology buzzword, and it is at the heart of IoT. First of all, let’s consider the amount of data that automotive telematics devices are expected to generate. A telematics device will produce a data record every second. This data record will include information such as date, time, speed, longitude, latitude, acceleration or deceleration, cumulative mileage and fuel consumption. Depending on the frequency and length of the trips, these data records or data sets can represent up to 1 GB of data per day, per vehicle!
To store this data, many insurance companies use distributed processing technologies such as the Hadoop file system. Hadoop is an open-source software framework for running applications on a large cluster of commodity hardware. Since Hadoop runs on commodity hardware that scales out easily and quickly, organizations are now able to store and archive a lot more data at a much lower cost.
To help insurance companies address the challenges from large data volumes generated by IoT programs, it is essential for insurers to implement an enterprise data management strategy. This data management strategy should provide a unified environment of solutions, tools, methodologies and workflows for managing telematics data as a core asset. It should also be flexible and scalable to reduce the time and effort required to filter, aggregate and structure the exponential growth in data.
With all this new data that’s available through IoT, how do insurers determine which rating factors are predictive. For examples which data variables can forecast driving behavior, structural defaults or healthy living.
The challenge is how to filter the noise from the signal. Adding a new variable increases the number of data points and relationships exponentially. In a very simplistic model, if you are testing for relationships among any five variables, there are 10 two-way tests to run, shown in the equation (5x4)/2 = 10. If you double the number of variables to 10, you more than quadruple the number of relationships to test, shown by (10x9)/2 = 45. With IoT sensors adding dozens, if not hundreds, of new variables, this creates the potential to analyze millions of relationships.
That's big data analytics! The problem is that many of those relationships may be redundant or trivial, and hidden among them are the “real nuggets,” or "signals."
The science of extracting insight from data is constantly evolving. Tools are more readily available, and industries are beginning to invest in the technology that supports big data. By using data exploration and analytics, insurers will be able to rank and weigh hundreds of new variables to develop highly accurate pricing models.
Figure: An example of a correlation matrix showing variables for auto insurance. Click image to expand.
Insurance companies cannot rely on traditional data mining technology to analyze all of this new data. Due to the sheer size of the data generated by sensors and IoT, insurers must consider a distributed, in-memory environment to display the results of data exploration and analysis in a way that is meaningful but not overwhelming.
Exploiting the Internet of Things
The IoT has the power to transform many aspects of the insurance industry and deliver significant competitive advantages to early adopters. But with this great potential also comes complexity. Advanced high-performance analytics and big data tools can assist companies in overcoming the complexities, enabling them to reach the full potential of IoT as it grows from a trend to a must-have for all insurers.
Stuart Rose is global insurance marketing manager at SAS. He began his career as an actuary and has over 20 years experience in the insurance industry.