In their need for speed, insurers are disrupting their own evaluation processes, reports Fiona Harris.
Business problems in underwriting were converted by UTS and OnePath into machine-learning tasks to aid development of a new underwriting solution.
The very traditional practice of underwriting is undergoing a technology-driven transformation, welcome news to those who labelled it the “department of business prevention” in frustration with the pace and complexity of the practice.
Those days are now in the past as new technology, centered on better use of data, is significantly speeding up the process.
And while many insurers have opted for off-the-shelf software, ANZ OnePath decided inventing its own would be more effective.
OnePath believed innovating the underwriting process with new tools such as data analytics and predictive modelling, would make it a true point of difference in its service.
Life insurers' underwriting problem
Back in 2016 only a small proportion of the 60,000-odd life insurance applications received by OnePath each year had a streamlined experience, according to ANZ chief underwriter Peter Tilocca. (And that was generally the uncomplicated ones.)
The remaining 75 per cent of applications spent between two weeks and three months in the ‘suspense file’, awaiting corrections, clarifications and additional evidence, such as tax returns and doctors’ reports, before the underwriting team issued its assessment. And that had been the process since the middle of the 2000s.
According to ANZ head of life insurance Gerard Kerr, the fact underwriting had stayed still for more than a decade, revealed that insurers had stalled, even though technology and society had progressed in many other areas.
Historically, the underwriting sector has been cautious about embracing innovation. However, recently the sector has harnessed technology to push applications out of the ‘suspense file’ faster, particularly by extracting greater value from the data they collect.
OnePath’s starting point
“We wanted to set a new standard in underwriting, to address the ‘elephant in the room’ when it came to insurance, and to push through new boundaries to make insurance a better experience,” outlines Kerr.
Being able to offer an immediate decision on up to 75 per cent of applications submitted electronically was one of the goals of OnePath’s underwriting transformation program. Achieving this would significantly benefit financial advisers and their clients.
“Our goal with everything we’re doing is to make the process as simple and efficient as possible, with the net result that more cases can be put straight through,” Tilocca says.
The premise was fairly straightforward: in-depth analysis of data would create a more customised application and underwriting process, reducing the information applicants needed to provide.
And the cumbersome one-size-fits-all approach where all applicants were required to answer all questions, regardless of whether they’re relevant to their circumstance, could be left in the past.
Revolutionary power of big data
“One of the things we’re learning through this use of ‘big data’ is that demographics play a lot larger part in risk assessment than we’d previously thought,” Tilocca says.
Through technology, underwriters would assess risk in a completely different manner and in the future potentially calibrate premiums more finely, based on behavioural and buying patterns, as well as demographics.
In need of a partner to bring about such change, OnePath turned to universities at the leading edge of data analytics and predictive modelling. After contacting a number of them, the University of Technology, Sydney’s Advanced Analytics Institute stood out.
“From our research perspective, we thought OnePath’s 10-year data collection was sufficient to build a reliable risk engine that could significantly refine and automate the existing underwriting process,” says Guandong Xu, Associate Professor at the institute.
To provide and get to work on more than 10 years of OnePath data to UTS, OnePath was allocated space in the UTS Industry Hub – an innovation lab for intensive collaboration and problem solving.
“We didn’t have much experience in underwriting prior to this project,” says Xu. “However, we knew that life insurance was much more complicated than other types of insurance as it requires domain experts to assess every individual case.”
The two institutions began collaborating with a pilot project to visualise the OnePath underwriting dataset.
The UTS Data Arena is a 360-degree interactive data visualisation facility that played an important role in developing the OnePath underwriting engine.
From that point the business problems in underwriting were converted into machine-learning tasks.
“I had some problem statements that I discussed with UTS, the overarching one being which underwriting questions are of more value than others,” explains Tilocca. “From there we were able to begin to work to optimise the underwriting rules engine.”
Once the analysis was conducted, the team could effectively rebuild the risk engine, stripping out the data not required and using behavioural economics to better ask questions to source key data.
“By using behavioural economics, we could ask fewer questions and obtain better quality responses,” explains Tilocca.
With a more comprehensive data set, a data-driven machine learning model was then fitted to the data. This laid the foundations for predictive modelling.
Results: reinventing underwriting
“The risk engine can automatically tune the prediction results based on the importance of each question towards underwriting,” says Xu. “An important advantage of this data-driven approach is that it self-learns over time as more data is collected.”
The results of the project are a:
- more streamlined process for the customer
- 30 per cent reduction in the time it takes to complete a personal statement
- more flexible and easy-to-understand approach.
And the risk engine has exceeded the university’s expectations.
“We can use the risk engine for quality assurance which was not planned at the very beginning,” says Xu.
While adviser and customer feedback will be gathered on the new risk engine, initial comments are favourable.
Speed of application, the flexible flow of the personal statement and the succinct nature of the personal statement were standout features to early users.
Fiona Harris is the marketing and public relations leader for ANZ Wealth.