If you find yourself on the phone with a Humana customer service rep soon and you notice that he’s suddenly changing his tone or message, don’t be surprised: He may be getting instant feedback about your emotional state — from an artificial intelligence assistant.
The AI assistant provides insights based on hidden signals in the human voice and may tell the Humana employee that he is “speaking quickly” or that you, the customer, “sound tense.”
In the next year, use of the AI assistant will be standard with 2,000 customer reps, Humana said. The quick rollout resulted because of an initiative the insurer set in motion a few years ago to fast-track solutions to consumer problems.
“It’s hard to overstate the importance of customer emotion,” said Kevin Schwake, co-founder of the FastStart Lab, the rapid-to-market initiative Humana started about two years ago to identify “customer pain points” and to find solutions much faster than typical in giant corporations. The unit’s motto: “Be scrappy — not crappy.”
FastStart Lab turned to Boston-based Cogito Corp., which developed the AI assistant for Humana’s call centers. In addition to providing instant feedback on customers’ emotions, the assistant can gauge employees’ performance. That may help the employee change tactics when needed, while frequent warnings may alert a supervisor to ask whether the employee needs help.
The AI also “learns” about the Humana employees’ typical speaking patterns and may flag pitch changes, unusual pauses or increasing speech velocity.
Humana used to review calls days after they had occurred, which, company officials said, was not very helpful because whatever lessons one could draw arrived too late for the call in question and might not have been applicable to the next customer interaction.
Now the insurer receives feedback instantly — and from every caller, rather than just 1.5 percent of customers who complete surveys after the call.
The program also helps Humana to establish a better relationship with customers, to appeal to their emotions, officials said. People with a positive disposition toward a company are more likely to recommend it to others — and are more likely to pay for that company’s offerings.
For FastStart Lab, ‘persevere, iterate or pivot’
For each of the pain points, the FastStart unit identified possible solutions, made sure its underlying hypotheses were correct and then tested the solutions on a small scale before collecting data and expanding testing.
For example, for one of the pain points, the unit generated the assumption that Medicare members did not have smartphones. The veracity of that assumption was critical because if the FastStart Lab generated a solution that required access to a smartphone, and half the customer did not have smartphones, the solution would not work very well.
To figure out whether Medicare members have smartphones, the company looked at internal customer data to see if they had provided cellphone numbers. Humana employees also visited guidance centers and asked customers. The insurer even sent employees to the state fair to gather feedback — and found their initial assumption was wrong.
Once the team identifies a possible solution to a problem, it begins testing, first on a small scale, perhaps at just one location, and tracks success measures such as customer scores. If the tests indicate success, Humana scales up the testing to about 200 employees, adds locations and extends the testing period.
Positive results from small-scale tests allow the unit to continue with the probable solution, while negative results prompt it to make slight alterations or completely switch direction.
The unit has found that small-scale experiments typically line up with the results from expanded tests, which reinforces that the process is working, Schwake said.
The unit also learned quickly that some of its possible solutions sounded good — but did not work well in practice.
Jeremy Benge, a consumer experience consultant, said that the unit thought it might be cool to “ping” customers on their mobile devices to notify them when they were leaving their Humana network. While Humana employees thought that such a service would prove helpful, feedback quickly indicated that customers thought such a “solution” primarily was “creepy.” So the FastStart folks switched to another solution.
Previously, without gathering that customer feedback, Humana might have spent lots of time and money to develop a service in which customers had no interest.
Schwake said that it’s OK for the FastStart to make mistakes — but it has to fail quickly and on a small scale.
Humana declined to disclose its investment in the lab and the projects’ financial results.
The FastStart Lab also created AskHumana, an online search tool that provides better results than the site’s previous tool. For example, if a customer searched for “deductible,” the old search tool’s results might have included a link to a five-year-old PDF that was useless for the customer.
Humana can gauge the success rate of AskHumana by tracking how many people click on search results. The response rate went from near zero to exceed 80 percent in a few weeks, company officials said.
Humana hopes to eventually tailor the responses to individual customers. For example, if a customer searches for the word “deductible,” the company hopes to provide information not just about deductibles in general, but about the customer’s specific deductibles and how much she has left to meet the threshold.
The unit also created FastPath, a touchscreen that allows customers to navigate through an easy menu to find information about billing or where to get in-network help near their current location.
It’s part of Humana’s effort to move away from apps, which people may not want to have to download when they’re looking for information, and toward giving people multiple ways to interact with the company, said Rich McCreary, a strategic consultant in Humana’s Consumer Experience Center of Excellence.
Humana leaders also hope the FastStart Lab experience will propel employees who spent time with the group to apply at least some of its techniques in their new roles, to help the whole of Humana become more nimble.