Spotlight On: Telematics: Map Reading

Telematics scores – the metric that insurers generate from telematics devices to assess individual driver risk and to feed back on driving performance to policyholders – come in many guises. While it might not be obvious at the moment, these somewhat abstract numbers are likely to differentiate the winners from the losers in the telematics market in years to come.

In these early days of telematics in the UK motor insurance industry – where anecdotal evidence of the total number of active policies is less than 250 000 – competing products have typically used a small slice of trip information to compile vehicle scores. The need to keep consumer propositions simple for a new class of product and achieve speed to market have justifiably focused attention on details such as the provision of a telematics device, a consumer portal and keeping data transmission and analytics requirements manageable.

That is all well and good while the number of active companies in the market is relatively small and drivers – particularly in the younger age groups – are dazzled by the amount they can typically save compared with their traditional policies. But what of the longer term, when price competition within the telematics market becomes fiercer and customers become less self-selecting, making identification of risky driving behaviours that lead to claims more critical?

As in much of insurance, any price is only as good as the data on which it is based. So, as the telematics market matures, scores that are based on detailed telematics data are likely to be more profitable.

The most predictive of claims scores will be based on data that is collected on a second-by-second – or more frequent – basis. This will not only allow insurers to study behaviours very precisely, but will also provide clean data to avoid the errors that are common in event counter-based scores.

Tracking the same journey taken by a driver – based on data received every kilometre versus every second – reveals a very different picture, with a lot of potentially valuable contextual information about the journey lost in the less detailed view.

External data will also enhance the ability of insurers to put driving behaviours in context. For example, driving 70mph on a motorway on a sunny day is not the same as doing it on a smaller road or in inclement weather. Scores that help differentiate how drivers adjust to their static environment, such as road type and speed limits, and their dynamic environment, including weather and congestion, will be materially superior arbiters of risk.


Knowledge equals power
Claims data will be a valuable source for serious telematics players to calibrate the score to the insurance application. By mapping claims to the exact moment they occurred, insurers will be able to isolate behaviours that actually cause claims instead of identifying claims that are correlated with increased risk.

This will produce a score that is significantly more predictive and enable more meaningful feedback to reduce future claims. Insurers could also take this further by using multi-variate analysis to identify combinations of factors that would identify an individual’s driving patterns as higher risk that are not already captured by existing rating characteristics.

Using these techniques, we have found that scores can differentiate a loss ratio by over a factor of 10 from the best 10% of business to the worst 10%. When miles are rolled up to the policy level, the score is already able to differentiate significant blocks of business that are materially under and overcharged. That means companies that have moved or are moving in this direction have a significant advantage over companies with scores based on limited data or simpler analysis techniques.

Knowledge will almost certainly equal power when it comes to earning a profitable slice of the growing interest in telematics.

Duncan Anderson
Global leader of property and casualty pricing and product management at Towers Watson