It is becoming critical to normalize all of this data to be able to use it for insurance purposes, regardless of source. Consumer confidence in usage-based insurance (UBI), after all, hinges on the accuracy of the data recorded and being able to give something back to the customer in terms of discounts or other incentives.
We commented in a recent blog article on why filtering and validating driver data is vital for telematics, and the work we are doing in this area in the US, UK and globally, with vehicle manufacturers.
Judging from a recent BBC Radio 4 programme, there is still a gap to be filled in terms of consumer perceptions of their risk score as it relates to driving style, how the technology works behind the scenes, and how insurers communicate with customers.
‘New drivers could save more’. ‘Telematics could be the right move for you’. ‘New drivers save an average of £400’. ‘Young drivers can save up to £917 today’. These are some of the punchy marketing messages that insurers present to customers on telematics and Black Box technology.
But how exactly does the telematics device determine that a customer’s driving is safe? Should algorithms be more transparent? Should there be a standard way of calculating driver scores? And are the current crop of apps and online interfaces meeting customer expectations?
Telematics and data volatility
The radio show discussed some of these questions and it took the example of William, an 18-year old driver from Basingstoke, Hampshire. William’s driving score resulted in a £50 premium discount in the first month of his telematics policy, £25 in the second month, nil in the third and £16 in the fourth month.
William told the programme he felt this erratic score didn’t match up with his driving behaviour, which he tried to correct on an ongoing basis, based on the readings shown in his phone app. In fact, his insurer identified a faulty installation in the early period of the policy, something it admits it should have been quicker to identify. The company went ahead to improve its procedures so as to be more proactive in this area, to be able to go back and correct an incorrect driver score, for the period when incorrect readings were being received.
Consumers want to know more
Each insurer will have its own procedure for smoothing the graph of fluctuating data readings, identifying any faulty readings and correcting any of the affected journeys.
Professor David Last, an expert witness in court case cases involving telematics data, told the programme that there are also inconsistencies in how different insurance companies score different styles of driving: with some scoring harshly based on hard acceleration, whilst others may emphasise hard cornering or speed (related to local speed limit or local road conditions) in their algorithm. Quality of the data is important after all, as a poor risk score can lead to a customer paying more, or it can result in policy cancellation in some cases.
“The GPS doesn’t work perfectly everywhere, for example driving into a car park or anywhere there is no GPS signal,” commented Professor Last. He added that insurance companies differ in how they handle a loss of the data feed, when the technology is not able to get a good fix on the customer. Each insurance company has their own algorithm that can put a different emphasis on different aspects of driving.
Tim Shallcross, Head of Technical Policy, Institute of Advanced Motorists (IAM) told the BBC there ought to be a standard definition of “safe driving” that drivers can understand.
“Safety ought to be safety,” said Tim Shallcross. “There ought to be a style of driving that everyone accepts.”
We suspect there are other issues in play, when it comes to how insurers use the telematics data and consumer perceptions. No doubt there is an element of natural human bias, when it comes to owning up to our own shortcomings as drivers and road safety. Some 90% of motor vehicle crashes are caused at least in part by human error. This is a basis on which to accept that humans are not always correct in their judgement of the road conditions and safety.
We should recognise also that the quality of scoring and data collection has improved from the early use cases, and the technology will only continue improving. Some older systems used GPS location data for measuring cornering, for example, an imprecise measurement compared to accelerometer data.
Insurance companies differ in approaches to how they run their telematics programmes, with some choosing to go it alone with their own analytics capabilities, some opting for outside help with the consumer interface and others opting for help with just the data management and risk scoring – an aspect where we are frequently asked to help insurers, from the largest global insurers to the smaller, regional insurers and brokers.
It shouldn’t come as a surprise that insurers are secretive about their own algorithms and how they relate to driving style. It amounts to their own intellectual property and their attitude to risk for an individual driver, as well as across the whole of their insurance book (just as they take differing approaches to quoting a premium to a customer).
We could expect some convergence, over time, in risk scoring models. But there is a real sense in which it is competitive. It’s been said that scoring is 90% science and 10% art.
Motorists, on their part, can try out different devices and apps provided by insurers, before they decide which policy to go for. They can also question their insurer if they feel their own score is erratic or if something is wrong.
From our viewpoint, we question whether there could be a move to a common industry-accepted risk score, derived from telematics. This after all would defeat competition and differentiation of products in the market. But certainly insurers could do more to communicate about risk scoring, and what drivers can do to stay safe.
The ingredients of the score itself, the risk attributes, frequency and accuracy of the data pulses can only get better and better, running as they do over Bluetooth and a mix of other data standards in the vehicle. And they will take another huge leap forward with 5G connectivity arriving in around 2020, helping to reduce the unit infrastructure and configuration costs for the insurer.
It is these aspects of the telematics service – and our end-to-end capabilities in the value chain – where we at LexisNexis Risk Solutions feel we are uniquely placed, to drive data quality for insurers, taking the broadest possible view, from the biggest possible data sample (not just within a single insurer’s base), and a device-agnostic view of risk scoring around the world.
It’s understandable that consumers want to know more about their own risk score, how it is derived, and the technology behind it.
This article is care of http://blogs.lexisnexis.com
This article can be found here
