ANZ OnePath using AI and fuzzy logic to avoid ‘the dreaded other’

More gains from partnership with UTS' Advanced Analytics Institute

Applying for life insurance is a long and often frustrating process. Thousands of questions on seemingly every medical condition ever suffered – except yours. Or so it appears.

“We’ve had multiple occurrences where people answer no to all the [medical] questions, then they come to the ‘other’ box at the end and they’ll go – ‘oh yeah I’ve had X’. And that question is actually back there, but they didn’t understand it so they defaulted to 'other' and started writing chapter and verse about their medical condition,” explains ANZ OnePath’s chief underwriter Peter Tilocca.

That’s a pain for both the potential client and OnePath. Whenever answers are given free form, typically the application will require the scrutiny of an underwriter.

“As soon as an ‘other option’ is selected in the underwriting engine it has to shoot off to underwriters and then there’s a lot of manual work that goes on between the underwriter, the adviser and the client to complete it,” says head of insurance operations at ANZ Wealth, Jay Tutt.

Now, a rewrite of OnePath’s underwriting system, which launched yesterday, is helping the company minimise what it calls the “dreaded other”.

Developed with the University of Technology Sydney Advanced Analytics Institute (AAi), the new engine is helping OnePath better understand why people stumble over the questions being asked. And ask them better.

Elbow nudge

Unveiled back in January, OnePath's far-reaching partnership with the university is investigating how client behaviour modelling, text mining and natural language processing, as well as social and predictive analytics, can “add value in the insurance sector”.

The work – which has been ongoing for around two years – has already used advanced data analytics to slim down the huge number of questions asked in insurance application forms, by cutting out the ones that aren’t relevant to the client, depending on who they are.

Late last year a successful pilot looked at the tradespeople customer segment, and was able to reduce a section of the application process from 32 questions to just seven without "any degradation to claims".

More recently attention has turned to bringing “plain English” to questions around medical conditions.

“Refining the question sets and to help people complete their responses more accurately, we’ve introduced what we call a master catalog and search function in the engine,” Tilocca said at a launch event at UTS in Sydney yesterday.

“Essentially people might not recall what their medical ailment might be; it has that smart fuzzy logic that matches to it,” he added.

For example, an applicant might not know they suffer from lateral epicondylitis, but they do know they have tennis elbow, one of around 1000 ‘medical aliases’ that have been added to the engine.

“So they can write tennis elbow and the engine will pick up what it is and ask them the right questions thereafter,” Tilocca says.

The search functionality has also been added to questions around travel. Applicants can type in ‘Disneyland’ rather than the address of the theme park, and they are served up with “all the Disney locations across the globe, you pick which one it is and you’re pretty much done and dusted from that perspective,” says Tilocca.

“It’s a whole catalog and search functionality; if you’re not sure, just start typing the first few letters, even if you spell it incorrectly the engine will pick it up and guide you,” he added.

Ultimately, the improvements mean more accurate applications, and less of the unexpected if a claim is made.

“You don’t want to be picking up these problems at claim time because you can’t do anything at claim time,” Tutt said.

“The end game of all this is to give a client a policy in real time. That’s the end game. There’s a long way to get there when you’re managing risk obviously, but we can utilise this data to get at least a lot closer to that point in reality,” he added.

Assurance and Alexa

Other analyses on a decade's worth of OnePath data have sought to identify which underwriters award claims incorrectly, or make exclusions that shouldn't have been made most often, it was revealed in May. This helps the firm establish improved quality assurance insights across its 55 underwriters.

ANZ has also been exploring the possibility of applicants having a conversation with a chatbot or Alexa to complete forms, which are currently done over the phone, on paper or online.

Voice interface form filling is “definitely not off the table” Tutt said.

“That technology, in its application to a complex industry like insurance; there’s still a fair bit of work to get there, to get that accurate. But it is something we’re definitely looking at,” he added.

The AAi has worked with a number of financial institutions including Credit Suisse, Westpac, AMP and HCF. It has also worked with government agencies including the Australian Taxation Office and the New South Wales Ministry of Health.

In 2016 the institute worked with wealth management group Colonial First State to build advanced analytics models to observe complex relationship patterns between investors, employers and financial planners and build predictive models to help improve customer service.

Join the newsletter!

Or

Sign up to gain exclusive access to email subscriptions, event invitations, competitions, giveaways, and much more.

Membership is free, and your security and privacy remain protected. View our privacy policy before signing up.

Error: Please check your email address.

Tags UTSAIinsuranceANZdata analyticsIOOFAssurancemachine learningadvanced analyticsOnePathfuzzy logic

More about AdvancedAMPANZAssuranceColonial First StateCredit SuisseFirst StateTechnologyUniversity of Technology SydneyUTSWestpac

Show Comments
[]