Chatbots have grown in popularity over the past couple years, but does this technology have lasting utility in customer support?
One study found that 59% of Americans have interacted with a chatbot or would be willing to. And why not? On the surface, chatbots seem like an intuitive solution to customer issues that is both cheap and automatic. This logic has led industry giants like Uber, NBC and Facebook to adopt chatbots to resolve customer inquiries. Does this mean we should delegate all support issues to chatbots, pat ourselves on the back and sit back while robots handle customer service for us?
Not quite. Unfortunately, despite lofty goals of total automation, chatbots are simply not equipped to take the place of humans.
The Importance of Context in Customer Support
The expectation of chatbot performance far exceeds their practical application. Ideally, a customer can engage with text-based AI, forgo a phone call and resolve an issue with the ease of a couple keystrokes. However, are there any metrics to support this technology works?
Actually, no. In fact, recent studies of chatbot efficiency actually reveal major flaws in the technology. Notably, this year Facebook doubled down on chatbot integration for their Messenger app and saw abysmal results: a staggering 70% fail rate for customer inquiries. Companies like Google, Microsoft, Kik and many others also suffered from low chatbot efficiency.
Yet, the fad continues to survive. Industry leaders are likely seeing the prevalence of this technology and are diving in uninformed. Let Facebook stand as a testament to the woes of chatbot early-adoption, because this premature technological shift comes at a huge cost to the reliability of customer support systems.
What we have learned from the mistakes of industry giants is chatbots cannot be a siloed strategy. Businesses are trying to move customers away from call centers and toward low cost, asynchronous channels such as email, SMS, text, webforms. The expectation is customers will get an accurate and personalized response from a chatbot, but this is simply not the case.
Chatbots lack the context needed to navigate customer inquiries on their own. Chatbots provide impersonal links that derive from a knowledge base, and this rigid approach to customer support does not account for the particularities of each inquiry. Without understanding of context, chatbots are left to provide canned responses that don’t get to the heart of individual customer need and often deflect more inquiries than they resolve.
This ineffective process usually leads customers to give up and dial a call center — the action the chatbot was meant to mitigate in the first place. Or worse, lead them to give up altogether and turn to a competitor. In order for chatbots to truly reach their full potential, they need to belong to a network of channels in which they can learn and provide a holistic support experience. Chatbots must be hinged to a conversational AI in order to regularly gain resolution context and adapt to customer need — giving customers a seamless experience across channels.
The First Step Toward Automation is Augmentation
Chatbots lack the fundamental interpretive capability that is so crucial to the customer experience. Customer support systems are meant to lend empathy and tailored resolutions to inquiries, but this is impossible with a chatbot. At the end of the day, chatbots are not ready to usurp agents because they are missing the human element, which is essential to navigating the nuances of customer inquiries.
Companies that adopt chatbots into their support repertoire have the right idea, but they are ultimately missing the mark. Customer support systems are ready for automation, but they must walk before they run.
Augmenting support agents with AI technology is the solution to ineffectual chatbots. The perfect compromise between system efficiency and customer satisfaction is to pair the human touch of an agent with automation. For example, natural language understanding technology automatically develops an awareness of how tickets have been resolved in the past and offers agents top recommendations on how to respond. Also, AI provides an overview of content such as macros and templates that are most likely to generate the greatest satisfaction score.
Automation is most effective when working alongside an agent and drawing from a volume of tickets that has a record of customer success. Thanks to machine learning technology, a type of AI that can teach itself, support systems can easily supplement agent interactions with automation. In this model, human agents have the final say on inquiries, and every resolved ticket trains the system over time to improve its accuracy.
Chatbots might not be ready to take the place of human customer support agents, but that doesn’t mean automation doesn’t already have its place in the industry. Augmenting customer engagement with automated processes like routing and recommended responses can streamline the support experience. As long as automation is aligned with a record of proven, successful resolutions, and works in conjunction with human agents, customer support systems will undoubtedly experience peak customer satisfaction through AI.
Pradeep Rathinam is the Co-Founder and CEO of AnsweriQ