Driving Change with Telematics

Andy Goldby, Chief Product Officer of The Floow, explains how insurance is changing with advancements in telematics and some of the various benefits it can bring to the industry.

When the first motor policy was written back in the early 1900s, no guidelines existed for an industry more accustomed to insuring cargo ships than cars. A lot has changed since, with UK motor insurance spanning 20 million UK households and paying out £27m in claims every day.

As we head full throttle into the 4th industrial revolution – the data revolution – the industry is undergoing an unprecedented digital transformation, which some will find intimidating. But harnessing and mining the vast lake of IoT data delivered by telematics offers a wealth of potential new insights. Whilst this is a demanding task, the data derived from telematics provides actuaries with powerful new capabilities to price their policies and inform their understanding of risk. Traditional factors associated with age, postcode, profession and claims history are no longer the only indicators. Telematics introduces a new set of policy-holder scores based upon a true picture of their driving behaviour.

Telematics Scoring Predicts Risk

At The Floow, we work with insurers across the world to deliver telematics solutions that are device-agnostic, enabling us to take journey data from any type of device, including smartphones, which are powerful mobility sensors.

We capture a variety of data. Firstly, sensor data which looks at GPS location, speed and phone usage and secondly, contextual data which records which roads are being driven, third party transient data and other external benchmarks. We translate raw data into hundreds of KPIs for every journey. For example, we know that though distraction can be caused by many factors, perhaps the most significant is using a mobile phone. There is a clear relationship between having a high distraction score and your propensity to claim – distracted drivers are twice as likely to have an accident - so our scores relate to the time spent on a call as well as handling a phone whilst driving. Similarly, in terms of contextual data, we have developed an understanding of the unique attributes of individual roads, analysing data in context of external factors such as how other people are driving, road layouts and pedestrian crossings.

Through various machine learning techniques data scientists then analyse and score against six key components of our algorithm, which include speed, distraction, smoothness of driving, time of day, fatigue and road risk. By blending a set of behavioural scores as well as contextual scores, we’ve developed a scoring platform that is proven to challenge the traditional proxy-based model which assess risk for an insurance policy. Our data gathered via client deployments over the past six years suggests that telematics portfolios deliver a 25% improvement in burn cost. Our scores deliver a predictive power to create up 10x difference in the likely claims frequency between drivers achieving a low score of under 30 and a good score of over 80.

Using this telematics scoring system alongside traditional rating factors (customer factors, vehicle factors and policy factors) can add significant value to the combined model’s predictive power, in fact adding 5-10x the additional impact that adding credit scores did.

When client actuaries and underwriters work with us to further train the scores against their own claims data, they report a significant boost of up to three times the profitability per telematics customer versus their traditional policy types, and a set of scores that represent powerful and unique new IP.

I believe that this presents a new paradigm of fairness into motor insurance pricing, and many senior decision-makers in the industry agree. In a YouGov survey we recently conducted amongst decision makers from international insurance companies, the widespread adoption of telematics is now seen to be dominant in shaping the future of car insurance. When asked how the motor insurance industry would change over the next decade, all the top five factors mentioned relate directly to the application of telematics.

YouGov Research conducted for The Floow (March 2018)

Autonomous cars will have required insurance companies to totally rethink risk 43%
There will be greater car sharing, replacing outright car ownership, and a move to usage based insurance 37%
Telematics will become the new benchmark for defining risk and pricing of policies 37%
Drivers will have an overall score for their driving ability, affecting their premiums 35%
The insurance industry will focus more on 'mobility' as an issue 34%

Identifying and Predicting Fraud

Fraudulent claims cost the industry many billions every year. The ABI states that insurers detected 67,000 fraudulent claims valued at £837m last year but it’s likely that many more went undetected, driving up the cost of motor insurance for everyone and impacting the profitability of insurers.

My view is that telematics offers very useful indicators to help insurers identify dishonest drivers. As an industry we will never be able to completely prevent fraud but understanding potentially fraudulent behaviour before it impacts your bottom line helps improve the likelihood of us being able to prevent the damage it could cause.

We have created indices that help to detect how honestly drivers are using our telematics apps e.g. the patterns of how they tag completed journeys and a measure of the contiguousness of the journeys tagged where the policy-holder is driving (they may appear to have missing journeys or appear to be tagging low-scoring journeys as if they were a passenger). Additionally, we have indices relating to the integrity of their declarations for mileage totals completed in the insurance year, or the ‘risk address’ where the vehicle is really kept and parked overnight. With these we are able to identify discrepancies which help to identify dishonesty within the early days of the policy enabling insurers to make informed decisions on how to treat these customers before their “cooling-off” period is over.

The Power of Social Science to Reduce Accidents

But it isn’t just about predicting risk. Important to our relationship with insurers is how we apply insights from social science, as well as the use of technology, to help improve driver behaviour, identify risk and develop an engaging claims proposition.

One programme we’ve found to be very effective in improving an individual’s score, is FloowCoach. We target drivers with the lowest scores and invite them to participate in a 12-week programme of telephone-based conversations with highly-trained behavioural coaches. Our latest research shows that for every 100 people in the lowest-scoring decile who have completed the programme, 16 accidents are avoided compared to standard feedback mechanisms.

Another way of engaging with customers is to offer them rewards and we have found that even small rewards can deliver significantly higher engagement and improvements in driving scores compared to the offer of an insurance discount at the end of the year – especially when the starting scores were poor.

Graph from our research showing the percentage increase in score improvement with rewards versus standard insurance discounts

To the Future

The other area which telematics will enable is the introduction of autonomy and driverless vehicles. When a machine is programmed to make decisions in place of a human driver, who is to blame when something goes wrong? This is of critical importance to the UK motor insurance industry where classically it’s the driver, not the vehicle, who is insured. When a machine is in the driver’s seat it will be critically important to fully understand the cause when an accident occurs. This kind of insight is only possible via the data that telematics can deliver, which is why we’re collaborating with several industry partners including Direct Line Group in the MOVE_UK project.

It’s our belief that telematics will be critical to the evolution of the mobility and insurance industries. Its applications are many, but what is clear is its insights will be a powerful contributor making mobility safer, and insurance smarter and more profitable.

This article first appeared in the June 2018 edition of The Actuary Magazine

Get in touch