There’s many opinions today on chatbots as they relate to customer service. Some point to them taking jobs. Others believe they will fail because customers prefer speaking with people. And some claim there are customer segments that outright hate them. The reality is they are here to stay and are predicted to grow in use.
The concerns raised about chatbots aren’t new. Change always brings cocern and debate. The automobile assembly line pioneered by Ford Motor Company (inspired by bakeries, canneries, and other businesses that used them at the time) celebrated its 100th birthday in 2013. Like any innovation, it brought both benefits and challenges. As we’ve seen over its time, the assembly line that started with human workers has been supplemented with robots and other automation that performs the mundane work.
Similarly, chatbots are just another form of job automation; here, that job is solving a customer’s problem. It’s hard to believe they’ve been around for over 50 years. In terms of real-world usefulness, though, they are still in their infancy. Not unlike the assembly line, they will continue to evolve and improve. When you consider how they make a difference for customer service, there are three key ways in which even the most rudimentary chatbots are improving customer service experiences.
Bridging the search gap
Twenty years ago I worked at a company that invested in knowledge management to replace an aging faxback system. Knowledge management promised a better way to manage information for both internal and external consumption and to connect problems to solutions.
The issue at that time (which still lingers to some degree today) is search. For search to work, the right set of terms are needed to produce a small and meaningful set of possible solutions. Therein lies the challenge. Customers don’t always know the rights terms or use terms that don’t match the scenario. As a result, when customer search a knowledge base, the subsequent list of possible answers can be overwhelming and littered with incorrect options.
Search has definitely improved over time. Documents are better indexed and can include hidden tags of synonyms of common search terms for particular solutions. Users can filter long lists of results. The most likely or most viewed solutions are pushed to the top. Still, the circumstances exist for one or more failed searches to end in a customer’s disappointment, resulting in a dead-end and a phone call to customer service.
Chatbots effectively replace search as a guide to finding solutions in knowledge articles. Chatbots can tap into customer demographics–what products or services they own, for example–questions the customer to narrow things down to possible solutions. That helps the customer arrive at a more likely answer, all in a conversational manner without requiring repeatedly frustrating searches offering multiple pages of the wrong solution.
Triage and handoff
Chatbots are not able to address every problem, though. The state of the technology today is to interact with the customer in order to collect details necessary to direct customers to existing solutions, which might exist in knowledge base articles, FAQs, automated solutions, or other places. For this to occur, conversation scripting has occurred behind-the-scenes. The script facilitates identifying problem symptoms from keywords entered by the customer through questioning, then the chatbot offers one or more appropriate solutions. In other words, chatbots do not troubleshoot or solve problems that don’t have an existing solution. They can ask the appropriate questions and collect details. So what happens when they don’t have the answer?
Enter customer service agent stage right.
Better chatbot systems allow a conversation to be seamlessly transferred to a live chat agent. That transfer includes all the information collected from the interaction: customer responses, solutions suggested, etc. The agent can then take over and work towards a solution.
The virtual cubemate
The two prior examples illustrate how a customer-facing chatbot makes service more effortless for the customer. In the contact center, it’s typical to have an internal “second level” or helpdesk for those issues that stymie the frontline agents. That same chatbot (or another that includes access to internal-only information) can similarly aid the agent. This can happen in two ways.
First, newer agents can become productive faster. Often customer service is structured with some agents taking certain types of calls or answering easier questions when they first start. A chatbot can assist them with answers to those questions.
But even seasoned agents don’t always have all the answers. They might not be aware of solutions to new problems that have cropped up. They also might not always recall every answer. They can have a quick consultation with a chatbot, which never forgets.
Ultimately, the customer also benefits. Though they did not use the chatbot themselves, it is assisting the agent with providing faster answers.
What does the future hold?
The automobile assembly line fundamentally changed an industry. When first implemented, what once took twelve hours was cut to two-and-a-half hours. Over time, it has continued to experience both revolutionary and evolutionary enhancements and now it’s possible to produce an astounding two cars per minute.
Like all technology, chatbots will continue to improve. The next big jump will come as chatbots take advantage of machine learning. Then, not only would chatbots learn the solutions to problems themselves without scripting, they would also have the means of solving new problems. Empathetic responses from chatbots is another big advancement and it might arrive sooner than we expect. And because this technology shares DNA with the voice-driven virtual assistants on smartphones, there is an abundance of usage data to advance the technology. With already half-a-century of development behind them, the pace of chatbot advancement is now poised to take off.