© 2019 SourceMedia. All rights reserved.

2018 will be the year of the chatbot

Given its large workforce requirements—U.S. companies employ an estimated 3 million contact center agents—customer service offers huge opportunities for automation. Yet chatbots, an artificial intelligence-based technology with enormous potential to reshape the customer experience, have thus far had very little impact on the contact center.

Despite the hype from some vendors, to date there are almost no examples of automated chats leading to reduced call volumes. But that’s about to change. 2018 will be the year when chatbots finally begin solving real customer service problems and making a significant contribution to corporate efforts aimed at digital transformation.

Almost all current customer care bots are limited to providing information. They can answer FAQs and respond to questions like “How do I fly with a pet?” or “How much does it cost to change my ticket?” by serving up links to articles and web pages. This is useful and, in some cases, deflects questions that would otherwise goe to a live agent.

Of much greater benefit, however, are the soon to arrive transactional bots that go beyond telling a customer how to do something, to actually doing something on behalf of the customer.

app tells.jpg
A user chats with Woebot, a chatbot aimed at helping users with anxiety and depression learn cognitive behavior therapy techniques in San Francisco, California, U.S., on Friday, July 7, 2017. . Photographer: David Paul Morris/Bloomberg

Developing such bots requires a considerably great investment and a higher order of development skills, since they must codify business rules and processes, integrate with backend systems and manage conversations. When a conversation moves from a simple question-and-answer to a complex multi-step transaction, the bot must deal with a greater range of user behaviors and undertake a much broader set of actions, including order confirmation, correction, clarification and pre-emption. Yet, despite the significant development effort, transactional bots generate an attractive ROI by automating a significant percentage of customer interactions.

Most businesses building chatbots today are starting from scratch. These companies often fail to realize, that a significant portion of their development effort frequently duplicates much of the work that went into their interactive voice-response systems, such as the design of conversational flows and the coding of business logic. In many instances, bot development teams repeat mistakes in interaction design, business rule implementation and usability testing that were made and corrected during a decade of IVR development.

Traditionally, IVRs and bots have been developed by separate groups from different organizations on incompatible technology architectures and platforms. In 2018, however, companies will finally realize that IVRs should be treated as speech bots, and that chatbots and speech bots are two sides of the same bot solution. Companies will begin to take steps to unify these development efforts, starting with the reuse of requirements, interaction designs, natural language models and backend middleware. This will lay a foundation for further innovation with bot tools and platforms that allow code and functions to be reused for both speech and chat experiences.

Human-Assisted Bots
Despite the claims of some proponents of artificial intelligence, human agents will never be entirely replaced by AI. Bot technology is imperfect, and there will always be circumstances when rules are unclear and human judgment or interpretation is required. The bot serves as a front door, automating the most common requests, while difficult, sensitive or low-volume requests are handled by the human agent. Over time, progressively more of the agent’s customer interactions can be automated, as AI models are trained on those conversations. By taking this approach, bot functionality can be continually improved, customer confidence in these applications will gradually grow and businesses will be increasingly inclined to adopt to adopt the technology.

So far, the AI wave has overpromised and under-delivered in terms of customer service. Existing bots have limited functionality and can only automate the simplest of customer journeys. But in 2018, bots will begin to deliver on their long-awaited potential, becoming an important channel for customer care.

For reprint and licensing requests for this article, click here.