Chatbots have hit peak hype right now. Need proof? Facebook Messenger had zero bots in February 2016 and 100,000 of them 14 months later. Brands are falling over themselves to explore the possibility of deploying some form of conversational self-service. They seem to want to put chatbots as the front ends for webchat, voice calls, and mobile messaging interactions. They want to build chatbot-like experiences for wherever Google Home and the Amazon Echo family of products belong.
There are chatbots to help you fight parking tickets, chatbots to help you book that glamorous vacation to the Amalfi Coast, and chatbots to keep insomniacs company through those long, sleepless nights. But companies have also deployed hundreds of chatbots to help with customer service. Travel companies such as Alaska Airlines, software companies such as Intuit, financial services companies such as USAA, and governmental entities such as the Enfield Council have all deployed chatbots specifically designed to help consumers resolve their own issues without assistance from human agents.
The list of companies will surely grow, and grow quickly. In my work as an analyst at Forrester, I am taking eight to 10 calls a week from companies looking for advice on customer service-focused chatbots. I’ve taken calls from companies in markets you'd expect (telcos, cable companies, insurance companies, retail banks, healthcare providers) and some that wouldn't normally be considered the usual suspects (a B2B electronic components distributor, the financing arms of auto manufacturers, and a B2B payment processing service provider). So, consumers need to get ready for an onslaught of customer service chatbots.
And companies need to prepare themselves for an onslaught of unhappy customers. The customers will be made unhappy not by the existence of customer service chatbots—consumers seem to want to self-serve, or at least they find self-service less painful than wading through the morass of phone-based, agent-assisted support. So, it is not the existence of the chatbots that causes distress; rather, it is the underlying company motivation translated into the chatbots' design and function that will make customers dislike them.
What is that underlying motivation? Deflection of interactions away from more expensive agents to lower-cost chatbots. That motivation should sound familiar to anyone who lived through the hype around interactive voice response systems (IVRs) as a way to contain customers in self-service. That company motivation for IVRs did not lead to happy customers, and the same will be true for chatbots.
When cost containment acts as the primary motivation behind chatbots, customer experiences often get pushed to the side. It was that thinking that led to "IVR hell," voice self-service systems that had no offramp and trapped customers in endless loops. Chatbots, and all AI technologies really, are still in the gawky, adolescent phase. They will get things wrong. They will misunderstand customers' inputs. They will assign the wrong intent to customers' words, when and if they do understand them. Companies need to analyze their customers' tolerance for getting it wrong.
But more important than moving away from a pure cost containment motivation is moving toward positive motivations, such as providing customers new or better experiences. Here's a quick example of what I mean:
A communications provider is using chatbots to solve simple, oft-repeated customer issues. The goal of the chatbot is not to tackle all service or even all types of customer service. The chatbot has clear limitations, but it is good at handling simple billing questions and service availability questions. This reduces the number of agents the company needs to staff its webchat channel. But rather than pocketing the savings, the company is taking those savings and using them to move its chat agents from outsourcer overseas back onshore and even plans to bring some of them back in house. So, the chatbot allows agents to work on the complex issues, and the savings from the chatbot is funding better-quality agents to handle those complex tasks.
Other uses for chatbots could be expanding the types of information available to consumers. One financial services company wants to be able to answer questions, such as, how much money did I spend with Disney last year? That is not the type of question I'd ever consider asking my credit card provider's contact center agents and expect a quick or consequential answer. The chatbot, then, could be a way to provide customers more meaningful experiences.
While chatbots can provide deflection away from expensive humans, that should not be the sole goal of your chatbot program. Customer experience still matters.
Ian Jacobs is a principal analyst at Forrester Research, covering application development and delivery.