Start Making Sense

If insurers could know with great accuracy which customers were likely to be involved in collisions and, thereby, increase their costs; better assess claims based on facts captured at the moment of the incident; or have the information necessary to create customized pricing tailored to the behaviours of individual customers, would they harness those opportunities?

Telematics makes this, and more, possible, but the masses of data collected through this technology are useless if the insurer has no way of analyzing it. In an industry that is frequently slow to adopt cutting-edge technologies, telematics is starting to make waves, and for those driving the adoption, will result in a boon to their bottom lines.

In the United States, Progressive Insurance was first out of the gate to implement the technology more than a decade ago and in 2012, the company reported that it wrote more than $1 billion in premium revenue for usage-based insurance policies. By 2020, it is forecast that more than a quarter of U.S. auto insurance premium revenue will be generated via telematics, representing more than $30 billion.

At the heart of telematics is data, but this is also the biggest challenge to companies’ successful implementation of the technology. Advanced analytics can overcome challenges presented by telematics related to making sense of all the data and, thereby, enabling companies to handle huge data volumes and weed out unimportant variables in order to identify important relationships. By removing these barriers, analytics can assist companies to quickly develop, test and use the best modelling techniques to create new auto insurance pricing models.

DATA, DATA EVERYWHERE

The value of data management to telematics

Big Data has become a technology buzz-term, and telematics epitomizes this trend. Experts forecast that insurers will need to collect at least 10,000 customer-years of data to be able to correlate driving behaviour with claims data so they can compare this information with data from standard drivers.

Telematics devices create massive amounts of data — approximately 1 terabyte for 100,000 customers each year — as the devices create a data record per second of elements, including date, time, speed, location, whether the vehicle is accelerating or decelerating, cumulative mileage and fuel consumption.

At the same time, bringing the varying data together into a single record is complex as it is usually comprised of many different formats. This can result in increases to cost, data quality issues from missing data, and complexity of bringing all the sources together for processing.

So before it is possible to start analyzing this data to extract insights, it is necessary to ensure that it is well-managed. Surprisingly, in talking to Canadian companies in all sectors, this step of data management is often neglected at a corporate level. In fact, a recent Canadian study by analyst group IDC and commissioned by SAS, found that an enterprise data strategy is not a priority of the C-suite. As it turns out, mid-level IT managers for Canadian firms are close to six times more likely than their international counterparts to be primarily responsible for a company’s data management strategy.

To make telematics work to the benefit of insurers, an enterprise-wide data management strategy should be adopted to ensure a unified environment of solutions, tools, methodologies and workflows for managing the telematics data as a core asset. To be successful, the strategy would incorporate data integration to improve the flow of accurate telematics information across the organization; data quality to ensure information integrity and excellence by managing the data quality lifecycle; manage the access and use of data across the enterprise; and master data management to create a single, accurate and unified view of all the telematics data.

How the data is managed will also help overcome another barrier to the adoption of telematics — consumer concern about privacy. Consumers want assurance that data about them will be used for the purpose stated, not, for instance, sold to the highest bidder for marketing purposes. By having a reliable s policy in place, it is easier to ensure customer information does not fall into hands that it should not.

NEEDLE IN THE HAYSTACK

From “unimportant” variables to important insights

It would be great to have a crystal ball that predicts which customers are risky. Failing that, however, using telematics paired with analytics might be the next best thing. For instance, using both can provide fast, fact-based answers to questions such as whether a driver who racks up 20,000 kilometres a year on highways is more or less likely to get into an accident than an occasional motorist who drives just 5,000 kilometres annually on city roads.

The combination of telematics and analytics can also help determine the most interesting variables to assess, thereby helping to solve the age-old dilemma, “I don’t know what I don’t know.” Using analytics and data exploration tools, insurance companies can sort through the masses of telematics data to determine which variables are important and which are not. For example, 20 years ago credit score was probably deemed an unimportant variable; now it is probably the most used variable in determining premium rates.

To ensure the tools used are powerful enough to analyze the volume of data produced by telematics, companies should look for solutions that incorporate a distributed, in-memory or high-performance environment to display

results of data exploration and analysis in a way that is meaningful. This technology makes it possible to quickly prepare, explore and model multiple scenarios using data volumes, which could not be handled by older technologies, to deliver accurate and rapid insights. This makes it possible to quickly trial several “what if” scenarios involving variables that may have once been deemed unimportant and develop models that can be quickly adjusted and re-run.

The ultimate promise of telematics is to provide auto insurers with increased competitiveness and profitability by making it possible to align individual risk and individual pricing using each customer’s specific driving behaviour. This would greatly reduce rate evasion resulting from customers misrepresenting information — either deliberately or not. This evasion is estimated to amount to 10% of premium revenue, according to numerous international studies.

STREAM IT, SCORE IT, STORE IT

Achieving this ideal is no simple task. As discussed, the mountain of data that is produced by telematics can be overwhelming. At the same time, the speed (or velocity) at which this data arrives is challenging.

To overcome this challenge, insurers can turn to high-performance analytics, which enable them to access and process varying velocities of data quickly. Insurance companies should consider a “stream it, score it, store it” approach. This enables analytics to be applied on the front end to identify the meaningful telematics data from the unimportant data or “the noise.”

Telematics has the power to transform the insurance industry and deliver significant competitive advantages to companies early to adopt the technology, but with this great potential also comes complexity. Advanced high-performance analytics and data management tools can assist companies in overcoming the complexities, thereby enabling them to reach the full potential of telematics as it grows from a trend in its infancy in Canada to one that is a must-have for all insurers.