We’ve all been there: waiting on hold as a recorded voice tells us that our business is important and an associate will be with us shortly. Nobody likes it, and the experience is often far from the usability and user experience goals the company had in mind when it designed the process. Why do we do this to customers? Hold messaging is required largely because of the relatively high cost of human operators. Up until now, call volume, call length, and number of agents have been the primary levers used to control costs, keeping calls short and agents succinct.
Of course, the possibility of using bots to field calls and other customer queries is attractive. Organizations have been using interactive voice response to route calls and handle simple requests for years now; chatbots can be seen as the next logical step, providing assistance when humans are not available, or when human help might not be needed at all.
Even better, the customers might just be thankful. Since bots are software, they can be scaled more affordably than humans, which can lead to faster service and even lower prices. Bots also respond faster and can be used across countless devices consumers keep at the ready, including smartphones, tablets, and old-fashioned telephones.
There is still a need for human customer service agents, but bots can play an important role if used them in the right circumstances and with user experience in mind. Enterprises should experiment with bots, especially to meet usability and user experience goals. When employing them within an organization, make sure to consider these best practices:
Do Meet Customers Where They Live
Everyone wants customers to have a good experience, so keep that goal in mind when designing bots. Don’t assume it’s not doable; in fact, when compared to the alternative, it may be possible to provide better service through bots. A great example of this is when customers have requests after hours and no human agents are available — but bots are.
KLM Royal Dutch Airlines provides its customers with an efficient chatbot experience that augments its customer support agents. The company increased brand engagement by going where its users already were. In addition to direct conversations, the airline offered user feedback via both WeChat and Facebook Messenger. Using APIs, they were able to tie these multiple experiences into a single interaction in their support dashboard.
Don’t Let the Bot Sound like a Bot
Chatbots can be human-powered, backed by artificial intelligence, or a mix of both. In each of these scenarios, remember that they must interact like a human. That means they should use friendly language that helps the customer feel valued. Of course, this isn’t an excuse to be unprofessional — the voice and tone the bot uses should match all other experiences customers will have.
If the organization is a financial institution, the bot is likely to use more formal language than a family restaurant’s bot would. But in neither case should bots sound like robots. Rather than programming the bot to say, “Please use your keypad to select from the following options,” consider something friendlier: “Hello there! Here are the ways I can help you!” There’s even a markup language called SSML that will allow developers to manipulate the characteristics of synthetic bot speech to make it sound more human. The customers will hear the difference.
Do Have Specific Usability and User Experience Goals for Bots
Most of us have heard the saying that you can’t be everything to everyone. The same truism applies to bots. Users will have a much better experience if they have an idea of what they want and need. As The New York Times found in its review of travel booking chatbots, specificity is important.
Expecting users to conform to the bot won’t be a good experience for them. Rather, look to employ the bot in situations where there is context regarding the customer and their needs. For example, if they’re looking through the self-service pages of an organization’s website, there should already be a pretty good idea in place regarding what they need. The bot could ask one or two additional pointed questions, and then be able to answer the majority of issues automatically.
Don’t Try to Have the Bot Do Everything
The concept of using chatbots in scenarios in which they’ll be most successful carries tremendous potential, but the promise of a general customer service bot is far ahead of today’s capabilities. Most artificial intelligence is simplistic, and while natural language processing is improving, bots cannot quite pass for human or truly understand what a customer is asking for just yet.
In the meantime, use bots in combination with human agents and other existing knowledge bases. Use a bot to route users to their answers, or to gather important context that makes their human interaction more efficient. In other words, find the situations bots work best in, and utilize them for only the specific tasks they’ve proven themselves capable of handling.
Do Continually Improve the Bot
You’re not going to provide a great bot experience for everyone, especially at first. To improve that experience and meet organizational goals, make sure to have the right data ready — analyzed for what is relevant to the bot’s performance and the customers’ needs — to make adjustments to your bot.
When an automated bot hands a customer off to a human, analyze the unsuccessful chat log and pair the “problem” with the customer’s eventual solution. This will help the team learn how to answer questions of that nature automatically, or at least be able to detect the sorts of queries the bot won’t be able to answer.
Survey the customer after a bot interaction to learn more about his or her experience. Just like other communication, keep this friendly and aimed at being helpful. Both quantitative data, such as rating the experience, and qualitative feedback, including open-ended questions, can be requested to get the information needed to make the user experience better next time.
Follow these usability suggestions to create a good user experience, see what’s working, and then begin expanding it throughout the organization. Both the customers and agents might be relieved.