Automation has become the Holy Grail of the customer experience domain. Since the rise of Interactive Voice Response systems in the mid-1980s through to today’s chatbots, the promise of automation has always been the same: keep customers away from talking to agents, reducing the need for agents, ultimately lowering costs. Vendors have (dubiously) added one more message—customers actually prefer an automated interaction—implying that if given the choice we’d rather use a chatbot instead of chatting with a human agent. (I personally don’t believe it and surveys validate my skepticism.)
Unfortunately the result of this technology enthusiasm is that if a customer does reach an agent, that is considered an automation failure. It is binary—either it’s automated (no agent) or not automated (live agent).
This is a deeply flawed view.
Customers talking to agents is not a failure of automation. But thinking that the sole use of automation is to keep customers away from agents is a failure of the imagination.
Automation is a rich capability that should be engaged throughout the customer journey and agent workflows to provide an effective and efficient experience that’s fantastic for all involved.
Agents are hired because we need their brains—to solve complex problems, to bring emotional empathy, and to positively represent the company brand. Despite what many vendors would have you believe, the state of technology is simply not good enough to replace all your agents. And it won’t be anytime soon. But that doesn’t mean we can’t use technology—and AI-driven automation in particular—to augment human agents and make them radically more productive.
Inspecting the work of a customer service agent highlights tremendous opportunities for automation. An agent’s day is filled with an assortment of workflows, processes, and tasks—looking up customer information, entering data, sharing documents with customers, collecting payment, typing contact summaries, and such. While it might seem simple to automate these efforts, that’s not the reality. Interactions are dynamic and context varies, so it’s nearly impossible for someone to write enough rules to intelligently invoke automation for all the possible scenarios.
While complex, dynamic and data-rich scenarios overwhelm the human brain, machines are perfectly capable of processing it all. Artificial intelligence technologies are designed to recognize patterns and meanings in what appear to be randomness—and through machine learning, to make predictions of the logical steps to take when a similar situation arises. When we use AI to drive the right automation at the right time—interspersing it with agent actions, we improve results in a way that’s not possible with a binary automaton-or-agent approach.
The economic impact of augmenting agents with automation can be massive. At ASAPP, we’ve seen customers dramatically increase their throughput, with one customer reporting an incredible 2x gain. That’s the economic equivalent of deflecting 50% of the calls away from agents—but easier to achieve, and more pleasing to customers.