Virtual agents and chatbots have become both commonplace and vital to ensuring customers are served quickly and efficiently through the channels of their choosing. And while today’s AI-infused chatbots are far more advanced than the decision tree-based, interactive voice response (IVR) automated attendants of yesteryear, care must be taken with their design and implementation to support an organization’s CX goals and initiatives, and to ensure customer success.
That is the message shared by Talkdesk’s senior conversation architect Dawn Harpster, who led a recent webinar focusing on the best practices for designing virtual agents and chatbots. Framed around a discussion of how customers are increasingly preferring to interact with virtual agents when they help accomplish tasks fast, more efficiently, and with little friction, Harpster identified several key design and deployment steps that should be addressed by internal teams and/or chatbot vendors.
Incorporate interaction analytics: Harpster says that the first step in designing a virtual agent or chatbot is to identify why their customers are contacting them, instead of assuming they know why. Reviewing previous calls, texts, chatbot conversation transcripts, email inquiries, and social sites can help identify the most frequent reasons, as well as the reasons that are best suited to be handled via a virtual agent.
Use simple and straightforward language: Customers should not need a Ph.D. to converse with a chatbot. Harpster says that virtual agents should avoid the use of jargon, and stick to simple and straightforward language. By adhering to this design principle, customers will be able to understand the chatbot’s prompts, and then respond quickly. The goal is to reduce the time it takes for a customer to get the information or task accomplished, and incorporating complex sentences or terms will have the opposite effect.
Do not obsess over your chatbot or virtual assistant’s personas: Customers do not need to believe that a virtual assistant or chatbot has a personality or human attributes. While the tone of the bot should match the organization’s brand, Harpster says there is little value in spending money on incorporating personality features, particularly at the expense of incorporating other functions. “The main focus is getting our customers and your users where they need to go,” Harpster says.
Ensure your intents are trained using your customer’s vernacular and phraseology: Customers may have wildly differing ways of asking a similar question, or referring to a particular product, service, or component. It is particularly important to make sure that when you are building your intents and your model that you are using training phrases that your users have generated. It is also wise to ensure that the model is trained on multiple variations of the phrase to account for regional dialects, age differences, and cultural differences.
Do not solely rely on menus: Harpster says that menus, where a virtual agent lists a series of choices, are rarely appropriate for chatbots, as they naturally limit the number of choices a customer has, and do not take advantage of the natural conversation style that customers use. Instead, she suggests designing conversation flows, which allow customers to organically request help in an open-ended manner, thereby allowing the customer to engage with the bot more fully, and provide more context about the specific issue they may be having.
Provide smart escapes: Another element to designing virtual agents and chatbots is to ensure the system can understand when customers are getting angry, frustrated, or their needs simply are not being met by an automated, virtual solution. There are situations where it makes more sense for the interaction to shift to a human, who has certain skills and capacity for displaying empathy that machines do not. Truly well-designed virtual agents should be able to detect, via the words being used by the customer, the volume and cadence of his or her voice, or even the number of attempts it is taking to accomplish a task, and then automatically push the interaction to a live person, without simply serving them with the “would you like to speak with a representative” prompt. Says Harpster: “Containment at the cost of user satisfaction is definitely not a win.”
Test Before Launch: Perhaps most importantly Harpster says that organizations must test their virtual agents and chatbots relentlessly prior to launch. This includes not only the front-end, customer interface, but also the APIs that are required on the back end. Additionally, it is important to have a range of testers across a range of ages and level of comfort with technology play around with the system prior to going live. “It’s always best to test with users who don’t know anything about whatever the chatbot or virtual assistant is doing,” Harpster says. “You can have the best designer in the world, you can design the best virtual assistant in the world, but when you start testing it with actual users, they’re going to do things that you, you didn’t plan on.”
The principles of good virtual agent design should mirror a company’s overall CX plan. There should be a focus on being customer centric, which includes reducing customer friction wherever possible; ensuring that the tools provided by the company make life simpler and faster for the customer; and making sure that customers’ feedback is accounted for, in terms of how they want to be serviced by virtual agents and humans.
Author Information
Keith Kirkpatrick is VP & Research Director, Enterprise Software & Digital Workflows for The Futurum Group. Keith has over 25 years of experience in research, marketing, and consulting-based fields.
He has authored in-depth reports and market forecast studies covering artificial intelligence, biometrics, data analytics, robotics, high performance computing, and quantum computing, with a specific focus on the use of these technologies within large enterprise organizations and SMBs. He has also established strong working relationships with the international technology vendor community and is a frequent speaker at industry conferences and events.
In his career as a financial and technology journalist he has written for national and trade publications, including BusinessWeek, CNBC.com, Investment Dealers’ Digest, The Red Herring, The Communications of the ACM, and Mobile Computing & Communications, among others.
He is a member of the Association of Independent Information Professionals (AIIP).
Keith holds dual Bachelor of Arts degrees in Magazine Journalism and Sociology from Syracuse University.
