Concept of ethernet cables plugged into a 3D rendered cloud

Want to Build a Chatbot? It Takes Technology and UX

19 Jul 2018 by Adam Deardurff

We've moved into the age where our time is at a premium and we're comfortable conversing through digital channels first. Chatbots provide a way to facilitate this conversation and provide value to both Business to Business (B2B) and Business to Consumer (B2C) companies. Your IT and User Experience (UX) teams have likely already started exploring what it might take to bring a chatbot to life.

To start this journey, it's often good to provide a Proof of Concept (POC) as a jumping-off point. At Insight, our approach delivers the ability to assess three key areas of need: technology viability, user acceptance and business value. Success, however, requires focusing your chatbot on a clear objective.

Give your chatbot purpose.

Bots have quite a bandwagon going, but they're ultimately still software. And that means long-term success is determined by providing a strong purpose. Failing to put thoughtful planning behind what your bot does and how it will interact with your users will inevitably end with your bot falling short of expectations.

Build an experience users need by first articulating a clear goal for your bot. What role will the bot play in your business? How will it add value? Where does it fit into the process?

Does success mean generating more leads? Reducing customer service time? Accelerating customer time to resolution? If you don't define success, your odds of hitting the mark are slim.

Map the user conversation.

Without natural human input, all chatbots might look like prescriptive wizards moving linearly through a process. Wizards, although useful, don’t realize the nature and power of chatbots. The reality is that users have their own way of navigating conversations, and most of the time, their preferences aren’t documented anywhere.

Spend time with users and the people within your organization who talk with users everyday, including sales people, customer service reps and technicians. These discussions will shed light on all of the key conversation points, allowing you to map the appropriate points that lead through the dialogue. You'll gain a better understanding of the primary path through your POC, as well as the potential divergent actions a user might take to reach the same end result.

Train your natural language processor.

You can allow your users to interact with your bot in a natural flow rather than having a scripted, robotic-like interaction (e.g., What is the weather outside?). But to do this, you need to use a natural language processor.

One of the most valuable features of Azure bots is their integration with Microsoft Cognitive Services and the natural Language Understanding Intelligence Service (LUIS) integration the service provides. By training LUIS with roughly 15 variations of what your users might say or request, the service can proactively identify user intent when a user deviates from your anticipated script.

Design the experience.

The term “design” typically implies using aesthetics to create your experience, but designing for chatbots requires planning beyond colors and fonts. Designing an experience for bots is about creating an intuitive and natural flow through the interface and bringing the brand attributes to life through conversational text. Users expect micro-interactions such as animation and real-time notifications to deliver the same type of responses as a live person reacting to a chat conversation.

Users might also be delighted to have some human personality injected so they don’t feel like they're typing back and forth with a computer. We call this the persona behind the bot. Does the voice and tone of your bot represent a real person, or do you make it obvious users are interacting with a machine? What's the voice and tone of the conversation? And, perhaps most importantly, how do you react to the unexpected?

Inevitably, users will test the boundaries of your bot interface, be it intentionally or accidentally. Having an appropriate response, providing contextual help and leading users back to the best path can make the difference between frustration and a great experience.

Prototype and test with stakeholders and users.

Whether the chatbot is embedded on your website or a stand-alone mobile app, a simulated experience or click path through the dialogue brings the concept to reality in a much shorter time frame. That means while you're developing the back end, you can be showing an artificial but more polished product to your audience.

Have users and stakeholders use the prototype and provide feedback. You'll get immediate validation on whether your chatbot delivers the value you intended. That also allows you to adjust course, if necessary, before you're heavily invested in integrating systems and forcing the chatbot upon your user base. Trust us, your users and business will thank you in the future.

Who needs a chatbot?

Regardless of your industry, there's probably an area for chatbots to bring efficiency to your employees, customers or services. Are your customer service representatives answering the same questions over and over? Do you have a sequential process for gathering information for quotes or troubleshooting? Do you need to prequalify individuals before passing them to a live representative?

Chatbots can save you time and money, improve time to resolution and drive increased sales.

Headshot of Adam Deardurff

About Adam Deardurff
UX Consultant

Adam is a seasoned UX practitioner focusing on digital strategy, research, analytics and UX architecture. He looks to apply his experience across the business and help expand Insight's UX offerings.

Read more posts by Adam
Connect with Adam on LinkedIn