As cost pressures mount across the insurance sector, brokers and underwriters are using new technologies to increase efficiency and improve outcomes.
As well as relieving cost issues, embracing technologies, such as automation, offers many opportunities for those investing in them.
“We already have the technology to automate the underwriting of an insurance policy from start to finish,” says Matt McGrillis, chief technology officer and co-founder of Send Technology Solutions.
“Across the market, we’re seeing that technology in action to varying degrees: initial checks and validation, data augmentation, pricing, and generating quotes can all be automated.”
High-volume lines, such as motor insurance, which use vast quantities of similar data that can be fed into machine-learning algorithms are obvious examples of automation.
Now more areas are embracing automation too. Lloyd’s of London recently welcomed Ki, the market’s first algorithmically driven syndicate.
Adding value through automation
There are two main forms of automation available to financial services companies.
‘Robotic process automation’ (RPA) executes simple repeatable tasks, while “intelligent automation” processes often make use of machine-learning tools to automate complex tasks.
Dave Ovenden, global pricing and underwriting leader at Willis Towers Watson, says the latter is usually “where you’re trying to add value”. While RPA technologies tend to provide greater accuracy and cost benefits, “intelligent automation gives you efficiency and effectiveness” he says.
RPA addresses basic issues, such as ensuring the same data is only entered once. This can help smooth out compatibility issues between systems that can result in duplication.
Intelligent systems, meanwhile, can facilitate forecasting.
“They allow you to make some sort of assessment about what is going to happen in the future,” explains James Willison, managing director at WCL, which provides services to insurance and reinsurance companies.
Automation in practice
Ian Macartney, head of innovation at Argo Group, says the Bermuda-based underwriter used artificial intelligence technology to streamline its data processes. The group previously engaged an outsourced vendor to input submissions but has replaced this process with AI-driven software.
“Automation helps to get proper submissions in front of underwriters a lot quicker so they can respond to brokers and then to the end customer in less time,” Macartney says. “This allows for better quality of submissions, as AI can prioritise the submissions as they come in and automatically decline any that we do not want to quote.”
The move to AI systems has helped Argo address cost and time lag issues that are associated with manual processes. The group is now using AI to read documents and accept or decline submissions based on pre-set conditions.
“We are automating the process of obtaining underwriting data, which not only helps to save time but gets more business through the door,” Macartney says. “In the future, with full digitisation, we can take a book of business that might have a high combined ratio and reduce the non-acquisition expense via automation. This will also help eliminate unprofitable business.”
WCL’s Willison says there is scope for further joined-up thinking of this kind. For example, he observes that “there isn’t automation from placing to accounting”, meaning the two sets of processes can become out of sync.
This can cause problems, such as unallocated cash. Errors can creep in because people have different interpretations of the placement.
“To me, a key part of automation, is to automate the accounting process from the end of the placement process,” says Willison.
This would create a “clear line of sight” from the risk that has been agreed to the premium that is received, he adds.
A data-centric model
Most interviewees say that RPA has advanced reasonably well, so now automation’s low hanging fruit is mostly picked. This means intelligent automation offers the widest scope for development and innovation.
To support this, data quality must be a priority.
“The full adoption of these processes largely depends on the systems carriers have in place, the quality of data they have access to, and, of course, the willingness to adopt and innovate within the business,” says Send’s McGrillis.
Increasingly powerful computers and algorithms can process huge amounts of data, and increasing kinds of data are being brought into the scope of algorithms.
As WCL’s Willison says: “The larger the data set, and the more computing power you can have, the more successful you will be in predicting future events.”
Adopting and adapting technologies and data sources requires investment in systems and expertise. Firms need to decide whether this necessary expertise can be developed in-house or will have to come from outside. Here again, companies may find it easier to buy rather than build, to allow staff to focus on their day jobs.
Buy or build?
While smaller companies sometimes find the cost of new technology to be disproportionately high there are no sign of smaller players in broking and underwriting being excluded from the automation revolution.
“We’re actually seeing the opposite,” says Send’s McGrillis. “The UK’s investment in insurtech is second only to the US right now. Companies of all sizes see the benefits of working with specialists to develop their technology and systems.”
He adds: “Though larger players may have the resources to establish specialist in-house teams, they are often working with outdated legacy systems that are not suitable for automation – at least not without additional system capabilities or modular systems built by insurtechs.”
Ovenden of Willis Towers Watson points out that some technology is priced on a “pay-per-use” basis or according to data volume, improves accessibility for lighter users.
A new world
While some roles are being lost as manual processes are automated, insurance industry experts believe a mass ‘deskilling’ is an overstated concern.
“The mix [of roles] will undoubtedly change,” says Ovenden. “You’ll need different people in different spaces.”
However, a range of potentially lucrative career opportunities are emerging for those who can combine traditional insurance skills in mathematics and actuarial science with knowledge of data science. Such people are “in short supply”, Ovenden says.
But Ovenden recognises that relationship management and client contact will remain vital skillsets for all insurance companies, as they cannot be automated and “people still do business with people they like to do business with”.
The Covid-19 pandemic has played its part in accelerating the adoption of automation.
As Send’s McGrillis says: “Without a doubt, Covid has forced a change management revolution. Two years ago, CIOs would have great ideas for how to drive innovation and embed technology, such as automation, but businesses were reluctant.
“The need to quickly adopt technology solutions has now been forced on many businesses and the C–Suite is asking what more can be done to push further forward.
“Covid has coincided with an explosion in data, and automation allows all of this unstructured data to be quickly structured and presented in new ways that make it easier to understand and work with.”
As machine learning and AI continue to advance, there will be more scope for insurers to streamline processes and improve turnarounds. For those yet to embrace such capabilities, there are significant opportunities to realise gains across the business by updating systems.