As befits an industry that is all about risk, insurance is inherently cautious.
It is getting to grips with the potential uses to which its reams of customer data can be put, and largely on the personal side. Even there, progress has been slow.
Too often, says Deloitte, a consultancy, the industry has taken a piecemeal approach to tech investment, transforming “system by system, function by function, and app by app”.
It reckons that advancing data management and analytics maturity should this year be a top agenda item for insurers.
Getting solid legal advice early in your company’s journey is invaluable
Water pollution has no one cause but many small steps and working together can bring great change
Empowering women in pharma: MSD Ireland’s commitment to supporting diverse leadership
Super nutritious, wildly versatile and oh, so tasty: Make potatoes your go-to food
Many “still too often treat data as an infrastructure expense to be managed, rather than as a strategic asset that can help them learn more about customer needs and preferences in terms of products and services,” it says.
Though slow, there has been some progress. Even before the pandemic the European Insurance and Occupational Pensions Authority (EIOPA), which reviewed the use of big data analytics within motor and health insurance, found that traditional data sources such as demographic or exposure data were being increasingly combined with new sources such as online or telematics data.
This was providing them with greater detail and frequency of information about consumers’ characteristics, behaviour and lifestyles, it said. That information enabled the development of tailored products and services and more accurate risk assessments.
The use of data outsourced from third-party data vendors, and the corresponding algorithms used to calculate credit scores, driving scores or claims scores, was under way.
Data analytics tools such as artificial intelligence or machine learning were being actively used by 31 per cent of firms surveyed, with another 24 per cent at a proof-of-concept stage. And that was before the digital acceleration brought about by the pandemic.
EIOPA reckons such trends offer opportunities for the industry and consumers. However, it cautions, there are risks too, particularly regarding ethical issues with the use of data, including its accuracy, transparency and auditability.
It raises concerns about what it calls the “explainability” of certain tools such as artificial intelligence and machine learning.
By now there are a number of data “use cases” in evidence in the insurance industry, says Prof Cal Muckley, chair in operational risk, banking and finance at UCD. These include situations where “thin file” customers, those with little paperwork, can procure insurance using proxy data.
The result is greater inclusivity, providing an opportunity to purchase insurance where someone previously might have been turned down. That benefits consumers and opens up new markets to the insurer.
Logging health or driver activity data via telematics can propel individuals towards healthier habits, as well as lower premiums, he says, while concerns about biases lurking in algorithms can be mitigated by use of a “uniform loss ratio” test – paving the way for individualised data-driven risk assessments. “The uniform loss ratio test can provide the guardrails”, says Muckley.
Data driven progress on the business-to-business side of insurance, however, is glacial by comparison.
“It is still pretty manual. We’ve lost paper but moved to Word and Excel,” says Hugo Wegbrans, global head of broking at WTW, the broking company formerly known as Willis Towers Watson.
On this side of the industry data is fragmented and rarely shared. It varies from – and remains siloed within – insurance companies. “The bigger the risk gets, the more individualised the data gathering is”, he says.
That’s because it is the data that the client provides – from how they do business to how they manage risk – that is the basis for the insurance. That tends to be heterogenous in the extreme.
The data given by the customer is then keyed in by brokers, before being rekeyed into the systems of, say, the five insurance companies that it requests quotes from. As well as being “hopelessly inefficient”, the margin for error is enormous.
“Everyone is looking to optimise this but so far no one has organised it globally,” says Wegbrans.
Insurtech solutions have successfully begun to disrupt personal lines but “it only works where the risks are homogenous,” he says.
If two apparently identical factories make steel railings, but one of them makes railings for industry and the other for balconies on high rise buildings, the risk is not the same, even if everything else is, he says.
“It is very hard to take a homogenous approach to industrial risk and if you get it wrong, you can get it horribly wrong,” he says. Risk errors over issues such as asbestos, for example, can result in claims dogging insurers for decades, and costing millions.
Yet the more data that can be gathered and shared, the better the advice the industry can offer clients, including suggestions about taking preventive action.
If, for example, a piece of kit has caught fire in one factory, the insurer or broker can encourage better maintenance of the same equipment in another. “The more clients are aware of risks the more they can prevent them,” says Webtrans.
It may be slow to arrive but such data sharing is coming. In April WTW announced that its broking platform can now enable digital trading with Zurich Commercial UK, as part of it’s continual digital evolution.
It allows the ability to exchange information quickly, efficiently and in an automated way, even for the most complex business.
“This streamlines the process to a real time transaction for both WTW and Zurich, representing a major step forward in modernising the industry,” says Wegbrans.
Slowly but surely even this, most cautious, side of the insurance industry is advancing towards a data driven future. “We are at the start of it,” he says.
