For ages, insurers have been crunching massive amounts of data to create the risk analysis profiles that help determine rates. The industry is no stranger to the value of data, so it’s no surprise that artificial intelligence (AI) is beginning to gain traction with insurers. As early as 2015, Tata Consultancy Services put the average investment in AI per insurer at $124 million. But is the industry adopting the technology fast enough?
With agile InsurTech companies like Lemonade nipping at the heels of legacy insurance companies and a potential shakeup from autonomous vehicles on the horizon, insurers must find new efficiencies, attract new customers, and provide a next-level customer experience for digital-age consumers.
While the insurance industry is beginning to use AI for everything from marketing, smart insurance products, market analysis, and fraud prevention, it has yet to scratch the surface with the capabilities of AI. “There’s a lot of lip service. There are pockets of really robust executive thinking around InsurTech and AI, but it’s also incredibly siloed. There also is a tremendous level of protectionism in place,” said Gregory Johnson, senior vice president of CCA Global Partners in a whitepaper by InsuranceNexus. In order to remain competitive, market leaders will have to speed adoption of new technologies to prevent disruption from upstart competitors and new business models.
Here are some of the most common ways insurance companies are using AI to better serve their customers and create new efficiencies and how they can potentially expand upon them.
AI-powered chatbots have reached a level of sophistication where they can carry on conversations in a very natural and human way. If you have used the chat feature on a website recently, you were likely chatting with a robot and didn’t even notice.
Natural language processing technology allows chatbots to understand and answer complex questions, like the questions that might be asked to complete an insurance quote. Though most of the industry’s current bots will still refer you to a human agent for that. GEICO’s “Kate” can answer fairly basic questions about your account balance and coverage, and help you retrieve documents in the GEICO Mobile app. Liberty Mutual has deployed Alexa Skills that can answer questions about its insurance, provide auto insurance estimates, and connect you to agents.
Allstate’s ABIe is an interesting use case for chatbots, as it is used internally to help walk insurance agents through selling commercial insurance products for the first time. Anticipating a decline in auto insurance policies due to autonomous cars, Allstate agents have begun selling commercial insurance, though most of the agents have no experience in that arena. ABIe (pronounced “Abbie,” short for Allstate Business Insurance expert) walks agents through the commercial selling process. It understands what product an agent is working on, where they are in the process, and who they are. It handles tens of thousands of inquiries a month — just imagine human trainers trying to deal with that!
Chatbots are capable of providing much more advanced customer assistance for insurance customers, and features like touchless claims service are indeed in the pipeline. And it appears that consumers are ready: according to a PointSource survey, preferences for using chatbots with insurance are no different from the preferences of engaging with retail companies. A total of 77 percent of consumers are okay with interacting with chatbots if it means avoiding wait times for customer service representatives with insurance companies, versus 75 percent for retail. However, there may be some resistance to do more with chatbots because the insurance industry model is based on using local, independent agents to write policies and assist customers. Insurers should rightfully be recalcitrant in removing that human touch, but more advanced use of chatbots could free agents to interact more with customers, build relationships, and make sure that the ever-changing needs of their customers are met.
‘Smart’ Insurance Products
The most common application for smart insurance products is usage-based insurance, which relies on telematics devices and data to assess driver behavior in order to determine the appropriate rate. In English, the little doo-dad you plug in under the dash or the app on the phone can tell whether or not you drive like a hooligan, track your mileage, see where and when you drive, and even if you have been in an accident. Your rate is then based on that information. For example, if you floor it off of every light, slam on the brakes, and powerslide around corners, don’t count on getting a good rate. But if you drive carefully, don’t drive too much, and avoid driving in big cities, you can probably expect a discount.
So instead of your rate being based on actuarial information, it’s based on how you drive. According to NAIC, this can increase affordability for lower-risk drivers, many of whom are also lower-income drivers. It also gives consumers the ability to control their premium costs by incenting them to reduce miles driven and adopt safer driving habits. Fewer miles and safer driving also aid in reducing accidents. The use of telematics also helps insurers more accurately estimate accident damages and reduce fraud by enabling them to analyze the driving data (such as hard braking, speed, and time) during an accident.
Advancements in AI will eventually help insurers use the data of individual drivers more efficiently and accurately to not only determine rates on an individual basis, but to predict the future insurance needs of customers to help agents provide a fully personalized experience for every customer.
Marketing & Customer Acquisition
While it’s possible that chatbots and AI can one day replace agents and underwriters, today many customers and insurers rely on the phone call to sell policies and assist customers. In the case of acquiring new customers, it takes a combination of online marketing and offline phone conversations to write new policies. Most customers will begin their journey researching insurance companies online, checking out reviews, and getting quotes. But when it comes time to write a policy, many conversions still happen on the phone.
The transition from online to offline poses a major hurdle for insurance marketers, because they can’t tell what happens once the customer picks up the phone. Did they get a quote? Did they sign a policy? Did they decide to keep shopping? What campaign or keyword drove the call? Without this information, marketers cannot accurately retarget and nurture people who have already called. They may be wasting ad dollars retargeting people who converted, or end up suppressing ads for someone who is still shopping. In addition, they can’t optimize their campaigns accurately because they can’t track what happened online that made the customer call. The solution for bridging this online-offline customer journey also lies in AI.
Call intelligence (sometimes referred to as call tracking) platforms like Invoca use machine learning algorithms to analyze live phone conversations to understand what the caller intends to do and what the outcome of the call is.
As these caller data points and outcomes are identified, a signal is automatically triggered in Invoca, providing real-time conversion and optimization data. These signals can then be pushed to the marketing stack (Google Analytics, AdWords, Adobe Experience Cloud, SFDC, etc.) in real time to make data-based marketing decisions. Marketers can utilize these insights to make smarter decisions on everything from PPC bidding strategy to retargeting.
In addition, this call data can be used to personalize the caller’s experience, which is especially critical when insurers use distributed call centers, whether that be local agents or regional call centers. Calls must be routed to the right agent in the right state and carry the customer’s information with it. Call intelligence solutions can not only properly route calls, but attribute conversions from those calls to the appropriate marketing campaigns. It’s a win-win-win for customer experience, marketing, and customer acquisition.
It’s estimated that insurance fraud costs the industry about $80 billion dollars a year in the U.S. alone. Consumers feel the pain from fraud in the form of higher rates and insurers get hit by the cost of sniffing it out and preventing fraud. There is a new sheriff in town to help police insurance fraud, and it’s artificial intelligence.
Insurance fraud runs the gamut from consumers intentionally causing property damage for profit to workers comp fraud and sophisticated schemes perpetrated by organized crime rings. In order to detect potentially fraudulent claims, AI-powered solutions can use data from past claims and known-fraudulent activity. IBM offers a Counter-Fraud Management for Insurance (CFM) solution is designed to help insurers prevent and intercept attempted fraud while detecting, identifying, and building the case against past fraudulent activity and improper payments. Powered by IBM Watson, CFM is able to help insurers detect fraud before it occurs, and more importantly, before they write a check.
Fraud prevention and detection is one of the most logical and potentially lucrative applications for AI in the insurance industry. It has the potential to create huge cost savings for insurers and consumers alike, and it is a function that currently uses an extraordinary amount of labor to accomplish. With AI helping human investigators work on cases, insurance fraud will become increasingly difficult for the bad guys to pull off and reduce one of the most burdensome costs that the industry faces.
AI and Insurance: Speed may be Key
It’s only natural that an industry that relies heavily on data to make financial decisions in so many areas of its business ends up using artificial intelligence solutions. The potential use cases in the insurance industry are manifold, but the speed with which insurers embrace the technology may determine their ability to compete in the near future.