Insurers Are Curious, Cautious When It Comes to Chatbots for Claims
The insurance industry is slowly recognizing the potential of chatbots to free up resources and improve customer service, even as it recognizes the potential of the technology to sour relationships with customers, according to a report released by research firm Novarica.
“Chatbots in Insurance: Overview and Prominent Providers,” profiles 13 technology vendors. Four offer chatbots specifically targeted to insurers; nine sell products that are designed for a broad variety of functions but can be put to use by carriers.
Jeff Goldberg, executive vice president of research and consulting for Novarica, said chatbots offer a low-cost way for insurers to expand the way they interact with their policyholders and agents. At the same time, carriers need to recognize the limits of the technology and be mindful of the stressful conditions under which many customers contact them.
“At this point, for most of insurers that are using chatbots, mostly it represents a very small percentage of their customer interactions,” Goldberg said. “When you dig in, you find that it’s a pilot that hasn’t been put to widespread use.”
Nonetheless, those pilots are being put to use not just by insurance tech startups, but also by established carriers, he said. Chatbots are often used for interaction with agents, but increasingly, they are moving into the claims sphere, specifically for first reports of loss and claims triage, according to the report.
Novarica said insurers are seeing chatbots as an opportunity for high-volume, low-transaction claims that require mostly routine data gathering. But the report warns insurers to be discerning about where the technology is used. Often, a consumer’s claim is triggered by a traumatic life event such as a car crash or fire.
Goldberg said most consumers may remember interactive voice protocols that were infamous for sending customers around in circles. He said some chatbots are equipped with a sentiment analysis function that will pass a customer along to a human if it detects that the person is upset.
Chatbots can be useful for customers who simply want to acquire information about the status of a claim, according to the report. Novarica noted that in previous research, it found that millennials prefer self-guided research and would prefer to talk to an agent only as a last resort.
The report profiled two instances in which carriers have put chatbots to use fielding customers’ claims inquiries.
Metromile, a pay per mile insurer, launched an app-based claims assistant called AVA that walks consumers through the process of filing a first notice of loss and filing an auto claim. The company said that 70% to 80% of its claims are approved immediately.
Merchants Insurance Group selected insuretch startup Hi Marley to allow customers to communicate with claims adjusters by sharing photos and receiving updates and notifications. The system is now live for all personal and commercial lines, the report said.
SmallTalk, a San Francisco-based technology provider, is among the 13 vendors profiled in the Novarica report. Its claim to fame was a customer acquisition chatbot for Next Insurance that allowed physical trainers and photographers to sign up for business insurance using Facebook Messenger.
That system, launched in May 2017, has since been taken off line because Next Insurance was concerned about “data handling,” said Josh Salwen, cofounder of SmallTalk and its executive vice president for technology.
Salwen said data management is one of the chief challenges facing chatbot technology. Businesses have trouble trusting the data that is gathered through chatbots. He said the insurance and financial services sectors are especially cautious because of privacy concerns.
But Salwen said he is confident that insurers will recognize the value of the technology.
“The one area where bots are good at, they know when they are stumped,” he said. “They will route you to a person. They tend not to give the wrong information. They give the right information, or they say, ‘I don’t know,'” he said.
“I would be shocked if eventually bots weren’t helping with almost everything.”