In a 2015 skit, Saturday Night Live depicts a customer service automation initiative gone wrong. The airline experiments with AI by introducing bionic flight attendants, depicted brilliantly by Scarlett Johansen and Vanessa Bayer, in the first-class cabin. Although initially wowed, the passengers’ mood quickly changes as the robots’ programmed services and responses fail to deliver. “Happy Halloween,” one robot wishes a bemused businessman despite it being the month of May (out-of-date content). The other mistakes a middle-aged woman as a scared child flying alone (faulty persona-identifying algorithm). Discontent ensues as the flight attendants start pelting everyone with hot towels (poor UE/UI studies and use cases). The Capitan, in an effort to reassure the passengers, unleashes praise upon the technology even if a software update is direly needed. At the end of the skit, for reasons known only by the machines’ programmers, one robot takes out the co-pilot and positions itself to fly the airplane.
As it happens, 2015 was a long time ago, and, in the real-world, advances in AI are looking much more promising. In NPR’s recent Planet Money newsletter, Greg Rosalsky reports on economists - led by Stanford University’s Erik Brynjolfsson - who studied a major company’s adoption of generative AI, Large Language Model, and chat and their impact on service rep engagement and customer experience.
Implemented a few years ago, the anonymous company’s 5,000 service reps communicate with their clients over chat and use a chatbot to answer questions and formulate customer messages. The researchers found that customer service reps were 14% more productive, happier, less stressed, faster, and less likely to quit. That is huge. In the customer service industry, 60% of reps quit each year. Customers also gave higher ratings to support staff and reported a better experience.
Some may see these results and muse about the end of human customer service reps. But this is not at all the key takeaway. The secret to this company’s successful AI-driven customer service initiative is in the details.
First, the chat was drawing upon a massive archive of years’ worth of customer/rep chat interactions. The chat’s success in providing accurate responses is based on the company’s own high performing reps. This insight is backed by findings. While the chat increased performance by 14% on average, there was little performance impact on the company’s high performers. The company knew how to exploit its customer service data to foster a positive experience.
Second, the chatbot freed reps to give more attention customer needs. The company saw an improvement in customer satisfaction, but this was orchestrated by the human rep. This company clearly invested heavily in AI but saw the chatbot as a tool for their existing workforce. Clearly such initiative was rolled out after careful consideration of customer needs.
Third, as we have seen, technology will continue to evolve. With advances in AI, chats will most likely become more effective in treating customer demands. The best companies will know how to leverage technology and adapt it to its current capabilities, assets, and target customers. This company under study did not just roll out hyped up technology but implemented it strategically.
What experience are you trying to achieve for your customer? What is your customer-facing teams’ role in delivering that experience? What aspects can be automated, and where does a human profile the best service? How can AI be used as a medium to exploit our data, knowledge, and capabilities? How can we implement digital to benefit both our employees and customers alike?
These results are amazing and worthy of praise. They point to the promise of AI, but such performance impact only comes for careful implementation.
The robot flight attendants are taking over the plane, and we laugh.