The situation
It was 2020, and a lot was going on. Few noticed that a regional bank in the northeast United States was about to onboard a new customer service employee, and that her new colleagues (there’s no other way to say it) were skeptical about her. Still, they needed the extra support and were willing to give her a chance. They pitched in to get her oriented, starting with the script that helped them help customers efficiently. Looking back now, she never thanked them for that. But it was OK: She wasn’t human.
She was a digital employee—a conversational artificial intelligence (CAI) bot, to be specific—so lifelike and responsive over the phone that it seemed she could think, interpret, and respond like her human counterparts. (She could, in fact, respond accurately despite the very diverse accents many of the bank’s customers had). Bank leadership had been exploring digital transformation with Deloitte for some time and brought her on to kick-start their “Everything Digital” vision, wherein clients could connect on any device, anywhere, anytime. CAI technology had proven it could help reduce mundane, repetitive tasks for retail call center employees, increase customer satisfaction, and shrink overall cost; a pilot program at the bank made good business sense.
(A side note about business value and call center customer satisfaction: For the business, time is money—the longer a customer is on the phone one-to-one with a support representative, the higher the cost. Time is also money for customers—they want their issues addressed as quickly as possible. But they also need to feel heard and not be rushed, or they’ll take their money elsewhere. A digital assistant helps in both directions; it’s able to work one-to-many, with no downside to customers lingering on calls as long as they leave happy. It can also work 24/7. This changes the calculus.)
A bank’s contact center was about to get inundated.
The solve
The bank’s new CAI agent was cloud-based software as a service (SaaS)—turnkey to both implement and upgrade. She (the voice was feminine) was deployed in the main customer service phone number so that she engaged before the interactive voice response (IVR) system, which sorts calls to various queues. She fielded all incoming client calls, handling the top five most common inquiries herself, and triaged the rest for handoff to IVR or a human colleague.
Deloitte built a proprietary analytics dashboard for the bank to understand and capture ROI on the CAI agent, as well as monitor and review call transcripts—for troubleshooting and resolving client issues, for analyzing conversational data, to enhance use cases, and to identify new ones. The dashboard also fed a monthly report to bank leadership with additional data related to ROI, including reduction in human agent touch points, reduction in calls going to contact center agents, and number of customers served during and after standard business hours.
Those top five questions—now addressed around the clock—accounted for 70% of inbound calls, considerably easing contact center agents’ loads. This was good because, remember, it was 2020: COVID exploded soon after the CAI agent hit its stride, shutting down brick-and-mortar bank branches and triggering an avalanche of entirely new customer calls. What about government announcements for pausing mortgage payments? Had their PPP loans arrived? What was their credit card status? Did their paychecks arrive (even though they hadn’t gone to work)?
You remember how it was: Every week brought new information and, with it, new questions. Yet the contact center was able to handle the surge in volume, addressing customer concerns without hiring new agents—this thanks largely to the CAI agent, which ramped up from fielding 32,000 calls per month to 100,000. Her colleagues, safely at home, were no longer skeptical.
That massive surge in contact center volume? Handled.
The impact
The original business case for investing in CAI had projected the bank’s newest digital employee would handle 800,000 calls over a five-year period. But with her shift to the front of the system, combined with the unforeseen volume of calls, she handled 1.1 million calls in just 10 months!
By the time the bank (and the rest of us) had weathered the height of the pandemic, she had successfully handled more than 2 million customer service requests, and the CAI agent was operating at a steady state, helping optimize both operational efficiency and customer experience.
By the time she logged 5 million conversations, she had resolved half of them directly and reduced human agent touch points by 40%.
This meant contact center employees had fewer simple and repetitive calls to field, improving service levels by freeing them up for higher-touch (and higher-value) customer exchanges. Customer satisfaction was up because they now had access to support 24/7.
Hold the line: conversational AI can lighten the load.
