(SoftBank’s Pepper robot mascot. Source: SoftBank.)

We’ve all been there.  You call the customer support line with the goal of speaking to an actual person, but what you get instead is a myriad of automated prompts, none of which leads you to a person or even a resolution. Frustration mounts and you try to hold back your anger—and the expletives.

In my particular case, I was driving through Connecticut very late one evening (I am not from Connecticut, but even I knew I was not one of the state’s nicer neighborhoods). I stopped for gas, but the only credit card I had with me was declined at the pump. I forgot to notify my bank that I was traveling and they put a freeze on my card for suspicious use. I didn’t feel safe hanging around the gas station alone, so I got back in my car and started driving assuming that by the time I found the next gas station I would have this straightened out. Things didn’t go at all as planned. I was almost immediately stranded in a voice-activated phone tree. My increasingly louder demands to “speak to a person” fell on deaf ears.  Running dangerously low on fuel and in a near panic by now, I let the first curse word fly. I barely had time to get the second (or maybe third) one out of my mouth, when I heard the sound of real person’s voice. And she seemed to know that I was agitated. It occurred to me that my cursing had immediately triggered the transfer to a customer service agent.  And she knew that I was being sent her way because I was upset, so she was prepared to handle my state of mind as well as my account issues.

Now, I knew this wasn’t some miracle of an artificially intelligent interactive voice response system (IVR). It was a smart programmer who recognized that a caller’s use of certain words means he or she is probably upset and that the call should be routed to a senior call center rep and flagged accordingly. But it is a step in the right direction—a step closer to considering the emotion of the customer when designing an automated system.

Adding Emotional Intelligence to Self-Service

Efforts to inject emotional intelligence into an automated self-serve experience—in this case via a simple algorithm—are just the beginning of a long journey toward the promise of an AI-powered future. In addition to curse-aware IVRs, we are also starting to see companies such as  Insurify attempt to humanize an automated self-serve experience via Evia “your expert virtual insurance agent.” Just text a picture of your license to Evia and, thanks to her connections with multiple data providers and carriers’ quoting platforms, she’ll text you back a quote in minutes. But Evia only works if the process is a simple one. She’s not equipped to handle emotion because the service she provides has a low emotional quotient—her customers have simple needs and a simple request: find the cheapest car insurance. You would need a very different version of Evia to effectively handle a post-accident call from a stressed-out customer with two young children in the background.

To see real innovations in the emotional intelligence of AI you have to look to Pepper. Developed for SoftBank by Aldebaran, a French robotics company, Pepper is a 4-foot tall, human-shaped robot.  He is loaded with cameras, microphones and sensors that enable him to gather data from his interactions, connect to the Internet and interact with any database or system.  What is most unique about Pepper is his ability to identify human emotions and select the behavior best suited to the situation and/or person. In one of many pilots, Pepper is playing a customer service role at banks in Asia and Europe.  Imagine walking into your local bank branch and being greeted by Pepper, who already identified your face and connected it with your account. He asks you the reason for your visit—perhaps you want to speak to the loan manager about a home equity line of credit—and while you wait to meet with the manager, Pepper begins gathering data relevant to your inquiry and assessing your needs both from a practical and emotional perspective.  By the time you sit down with the human loan manager, she is well-informed of your needs and can focus the conversation on reassuring any doubts you may have, talking through your options and helping you close the loan

Pepper and similar robotic technologies not only serve a functional customer service and sales need, they also break down preconceived notions about how human-machine interactions can work.  When only 10 percent of people in an Allianz Life Insurance study say they would be comfortable having their relationship with their financial advisor exist entirely online, creating more positive and productive automated experiences for customers is critical to moving the needle. Pepper and other technologies like him help make customers more open to and accepting of their use in a host of other situations.

Take the AI Train

All of these examples point us toward a pretty exciting future. Through them we can see machines becoming more like humans in their ability to read emotions, personalize interactions and customize problem solving. It’s no wonder so many insurers are clambering to jump on the AI train. But it’s worth tempering our excitement with a dose of reality—and not just the reality of the current technology limitations, but the reality that even with emotional AI a human will always be part of the equation. After all, insurance is a relationship business and our customers are human. If we mistake intelligence for empathy our machine-powered interactions will ring hollow, putting the relationships we’ve built with our customers at risk for the sake of innovation and the promise of efficiency.

So where does that leave us? Many insurers feel as though they are on a precipice of sorts as they try to figure out their strategy for investing in automated technologies—that they have to make a huge leap in order to gain a competitive advantage. There certainly will be some leaps to be made along the way, but before companies do any jumping, we encourage them to ask themselves the following questions:

Do you understand your customers? Do you know here they are in their lives—and not just from a demographic or market segmentation perspective?  What do they need from their insurance—a quick-hit auto policy or hand holding through a difficult stage of their life? It’s incredibly important that you understand your customers so you can put in place the approaches and resources (both human and machine-driven) to meet their needs and strengthen their loyalty.

Do you understand their end-to-end journey? Automation and self-service can only deliver on their potential if they are applied at the right time and in the right context for your customers’ needs. Mapping the customer journey is essential to identify the circumstances in which automation and self-service is sufficient or human intervention is required. Figure out what your customers are facing at any given point in their journey—the situational or generational drivers of their needs.  And understand the Emotional Quotient required for the meeting those needs.  A friendly, but efficient text interaction may suit the needs of a college grad seeking cheap auto insurance, but this will certainly not help a young family whose house has been destroyed in a hurricane.  If you don’t invest the time and effort to understand the customer journey, deploying even the most advanced technologies will have little impact.

Do you understand your agents and underwriters?  Agents and underwriters are invaluable to your business.  Agents are the bedrock of your customer relationships and underwriters ensure the relationship is mutually beneficial.  But do you really understand what it is they do and the attributes they possess that makes them so uniquely valuable? Do you know how much of their time is focused on the high-value relationship building work versus low-level administrative tasks?  It has been estimated that underwriters spend 70 percent of their time performing low-value tasks, such as searching, aggregating and selecting data, and only 30 percent of their time in risk selection. And insurance agents are dragged down by the demands of multiple systems, repetitive data entry and other time-consuming manual tasks. Understanding where your agents and underwriters fall on this spectrum of productivity enables you to pinpoint the technology investments that will maximize their impact and minimize the amount of their time that is wasted on low-value tasks.  These may not be the coolest technologies, but they’ll do wonders for your employee productivity and your customer satisfaction metrics.

Is your data ecosystem in order? Data quality is a key determining factor in the performance of robo-agents and robo-underwriters. In other words, if you are putting junk into the system, only junk will come out.  You may have the best agent in the business, but if she’s using bad data to tailor a product recommendation, she’ll produce bad recommendations and potentially alienate her customers. When clients feel as though their agent doesn’t know them (or worse, doesn’t care), it’s an uphill battle to win back their confidence. Avoid this predicament entirely by taking the issue of data quality seriously – both within your organization and your larger ecosystem. It’s one of the most important things you can do in pursuit of a successful robo-future.

Do you have your products clearly categorized based on level of complexity? Is your product portfolio aligned to customer need? Do you understanding which products are fit for delivery through self-service models versus those that require the guidance of an agent?  And are there some that can benefit from the influence of both?  Knowing your product offerings inside and out — and understanding how and when your customers interact with them —is at the core of your ability to personalize them, whether delivered by a machine or a person.

I appreciate that my credit card company had a mechanism in place to transfer me to a customer service agent once I started cursing, but what I appreciated even more was that the customer service rep was ready and able to help me. She had the information I had already provided the automated system and, more importantly, she had a calming demeanor that helped me recognize she was there to help and everything would be okay.  What will successfully power the industry’s robo-future and your piece of it is a truly thoughtful approach to your stakeholders’ needs and the products that serve them—along with the deployment of the right intelligence technologies. But it requires a lot of self-awareness as a company, a deep understanding of your customers and employees and a healthy dose of skepticism about all the pretty shiny technologies increasingly disguised as humans. Are you ready to look in the mirror?