Digital Insurance spoke with Sheila Anderson, CIO of Aflac. The Covid pandemic spurred the supplementary insurer to use
Aflac aims to get the most out of AI for customers and claims
May 8, 2023 6:16 AM
Photo by John David Helms
How has Aflac’s deployment of AI impacted its operations?
John David Helms
We have seen a direct correlation between the filing of a claim, the time to pay that claim and our customer staying with us. We have about 46% of our claims that are actually fully automated or what we call straight through processing, primarily in the area of wellness claims [which are rewards for regular checkups and preventive care].
About 49% of our manual claims volume today results in less complex claims with lower payout. That represents about 5% of our claims payouts. For all of these simple claims that don't require proof of loss, like our wellness claims, for example, our goal was to pay those as quickly as possible, so our customers can see very rapid value. We really don't want our specialists spending a lot of time on those. The goal is to free up specialists so that when they're dealing with claimants or policyholders that may have a more severe health-related issue, they can spend more time where insureds actually need to speak with someone in a more complex, challenging situation.
Now we're actually rolling that out across multiple lines of business, so it is moving to a point of scale. We're also beginning to see our expected business benefits and value.
How is Aflac implementing these uses for AI?
One area is the digital service center, focusing on our customer data and improving our overall customer service experience. We have a lot of internal data. We paired that with AI and machine learning to analyze that customer data in real time so that we can identify trends and then make more personalized recommendations. Those help improve that overall customer experience and overall operational efficiency.
AI/ML can implement sentiment analysis to proactively tell that a customer is showing signs that they may be choosing to cancel a policy, based on interactions in conversations. We focus on customer persistence. We want those policyholders and those customers to persist and remain customers at Aflac so that we can continue serving them and giving them the value that they need.
On the sales side, we elevated a guided selling tool that would assist customers in determining which products would best target and fit their needs. We did it on things such as a look-alike model, product attributes -- knowing that our agents who were very accustomed to being in person sitting across the desk doing that exchange with an individual, wouldn't be there. So we tried to mimic that desk experience by giving them those options.
I have a team that works across the organization and they engage very directly with our business to define those customer journeys. They do everything from the journey experience all the way through the research design, UX as well. We go past just reacting in our communication, really focusing on proactive communication, pre-emptive personalized engagement at scale.
AI/ML can implement sentiment analysis to proactively tell that a customer is showing signs that they may be choosing to cancel a policy, based on interactions in conversations. We focus on customer persistence. We want those policyholders and those customers to persist and remain customers at Aflac so that we can continue serving them and giving them the value that they need.
On the sales side, we elevated a guided selling tool that would assist customers in determining which products would best target and fit their needs. We did it on things such as a look-alike model, product attributes -- knowing that our agents who were very accustomed to being in person sitting across the desk doing that exchange with an individual, wouldn't be there. So we tried to mimic that desk experience by giving them those options.
I have a team that works across the organization and they engage very directly with our business to define those customer journeys. They do everything from the journey experience all the way through the research design, UX as well. We go past just reacting in our communication, really focusing on proactive communication, pre-emptive personalized engagement at scale.
What other functions is Aflac developing?
Our biggest focus in the next year is scaling the existing capability because we see great value. We're scaling some of our existing models to the larger ecosystem and across the other lines of business, including Japan. Our team is looking to drive value to our customers, starting with our enrollment experience. We're starting to segment pieces of our customer lifecycle journey where we can be more effective and efficient. We will have several of those stood up so that we can really focus on driving that value to improve our overall customer experience.
Have there been growing pains in scaling up these areas?
The growing pains are in the beginning, making sure you're focused on truly partnering with your customer in the right way in your business. We have an extremely exceptional partnership with our business. It's what I'll call business-led, IT-enabled, which is fantastic when you can get there with a business. We have a very strong partnership on the front end, of a team of data scientists that focus on the analytic side, they understand that data inside and out.
It's really getting that right mix between technology and enabling the business vision with the technology and then carrying the consumer side of this making sure they're a part of the journey, making sure that there's not a fear factor associated with it because there's so much hype. We must have some opportunities and ways to govern the data that goes into our models.
ChatGPT is a good one. How do we make sure that that experience is a solid experience to take that fear factor away? Bringing our customers along, bringing along those people that would be impacted and the business change that will happen because of the new types of solutions. You have to ensure that you're bringing your consumers along, and they understand how and why it's changing their business process, that it's actually adding value. Because most of those folks are probably used to doing a lot of manual activities. It's more of how do I trust this model, that it's actually telling me the right answer, and it's a valid solution.
It's really getting that right mix between technology and enabling the business vision with the technology and then carrying the consumer side of this making sure they're a part of the journey, making sure that there's not a fear factor associated with it because there's so much hype. We must have some opportunities and ways to govern the data that goes into our models.
For Aflac and the industry as a whole, what's the potential of AI?
Most people are more familiar with the property and casualty space. But what they really want to do is to buy that assurance and peace of mind knowing that their risk is covered. For insurers, we want to be able to do that profitably and as efficiently as possible while providing that customer experience.
In the next few years, those customer experiences will be where you see a lot of the focus – and on the operational efficiency side. Also looking at internal processes, whether it's back office or any place where you're doing a lot of manual activities that could be task oriented or automated. There's a prime opportunity for AI and machine learning.
We're looking at a process to assess use cases and make sure that you have a clear view of what value is expected. How do you evaluate those inside of your company to drive the value that you need in the future? It's really key to have a process where you're able to evaluate those effectively and make sure that you have a clear view of expected outcomes you're testing and learning. If models aren't giving you what you need, how do you revise them, stop or pivot?
In the next few years, those customer experiences will be where you see a lot of the focus – and on the operational efficiency side. Also looking at internal processes, whether it's back office or any place where you're doing a lot of manual activities that could be task oriented or automated. There's a prime opportunity for AI and machine learning.
We're looking at a process to assess use cases and make sure that you have a clear view of what value is expected. How do you evaluate those inside of your company to drive the value that you need in the future? It's really key to have a process where you're able to evaluate those effectively and make sure that you have a clear view of expected outcomes you're testing and learning. If models aren't giving you what you need, how do you revise them, stop or pivot?