Conversational Services

Building better chatbots: Two questions to ask before you get started

Share this post:

Key Points:
– Chatbots are the new apps, offering scalable, instantaneous, 24×7 interaction that is difficult to achieve with human agents.
– I built a chatbot that could answer questions about myself — an interactive, conversational resumé of sorts.
– Through trial and error, I found two important considerations to take into account when building a chatbot.

Learn to build a chatbot in minutes

 

Chatbots are becoming more and more common as channels for engagement. Whether they’re answering customer support questions, helping employees pick the right medical plan, or serving as a personal assistant, bots offer scalable instantaneous, 24/7 interaction that is very difficult to achieve with human agents.

I have seen businesses across all industries identify a wide variety of valuable use cases for their own conversational systems. But finding the right use case for a bot is just the beginning. Designing a system that can successfully engage users, whoever they might be, in genuine conversation, requires careful thought. And making sure that system is achieving the right goals —having the right conversations — is critical.

Recently, I set out to build a fully functional chatbot on my own using Watson Conversation Service, in an effort to learn how to do just that — create a conversational system that can drive conversations in my desired direction. For my project, I chose to create a chatbot that could answer question about a topic with which I’m very familiar: myself. I wanted to create an interactive, conversational resumé of sorts. Let’s call is “Susannah-Bot.” And through quite a bit of trial and error, I found that there are some unique considerations that we have to take into account when building a two-way, conversational experience in the form of a chatbot.

In particular, there are two questions that you need to ask if you want to build an effective bot:

1. What do you want people to ask your chatbot?

The first question that most people ask, when faced with the task of creating a chatbot, is, “What kinds of questions will my users ask this bot?” From a user experience design perspective, this is a completely valid first question. Anyone familiar with design thinking will tell you that this human-centric approach is the best way to create something that is actually useful to your target audience.

But with chatbots — or any cognitive system designed to interact with users in a human-like way — we have a unique opportunity to drive the digital conversation with our users instead of simply reacting to their inputs. We can create systems that, unlike traditional static web pages or applications, can constantly offer feedback to users and nudge them towards mutually beneficial behaviors.

And so, one of the most important questions to ask when you first set about creating a chatbot is, “What kinds of questions do I want people to ask?”

With Susannah-Bot, I wanted people to ask questions about my career, my experiences, my skills — the kinds of questions you might find the answers to in someone’s resumé. But when I first put my bot in front of a few guinea pig friends and colleagues, most of them started asking it questions like, “What’s Susannah’s favorite color?” I learned that I needed my bot to provide people with suggestions that would guide them to my intended line of questioning. Instead of a simple, “What would you like to know about Susannah,” my bot now introduces itself by framing the scope of its knowledge within the realm of my professional life.

Here’s an example of how asking the question, “What do I want people to ask this bot?” can have a huge impact on your business. One of the most valuable things a customer support agent can do is help customers get the most out of a company’s offerings. But most people come to a traditional support site or knowledge base with a very specific question in mind, like “How do I reset my password,” for instance.

With a properly configured chatbot, you have the opportunity to transform every one of these simple “password reset” interactions into a meaningful conversation that will make your customers more successful. You can configure your bot to follow-up basic answers with valuable tips or questions to learn more about that particular customer’s usage patterns and goals, to offer personalized help down the line.

One of the most powerful features of a conversational system is the ability to help users uncover answers to questions that they did not even know to ask. And in order to this, you need to think seriously about what those questions can and should be.

2. Who does your chatbot need to be in order to be successful?

Building trust with users has always been a critical part of the design process. But with conversational, cognitive systems, building trust is more important than ever. Our ability to drive meaningful conversations with chatbots relies on our users’ willingness not only to engage with but also to trust and listen to the system.

So, who does your chatbot need to be in order to build that trust with its users? Think about how your own expectations change depending on whom you’re talking to — the kinds of questions you would ask a sales associate are quite different from those you would ask a technical support specialist. If you want to build a conversational system that your users are going to trust to provide them with information, you need to set the right expectations by giving that system the right persona.

When making Susannah-Bot, I faced an interesting dilemma: should I make my bot a virtual version of myself? Or should I make it an omniscient third-party, a trusted source that knows everything about me? Ultimately, I decided to go with the third-party persona. I thought that people would be more likely to believe my chatbot if they felt the information was coming from an unbiased source. After all, anybody can make up facts about themselves on the Internet — why should anybody believe what “virtual Susannah” has to say about herself?

Once you’ve settled on the right “identity” for your bot, it should influence nearly every one of its functions. The way that it introduces itself to users, how it phrases answers, what it says when somebody asks it a question to which it doesn’t know the answer to — these are all opportunities to showcase that persona, and in turn, build trust. Just as novelists construct three-dimensional characters with nothing but language, you (or your copywriters) should embed personality into the responses and speech patterns of your chatbot.

Also fundamental to building trust is transparency. While you want your bot to have a firm identity, you also want people to know that they are interacting with a bot. Being upfront about this fact is one of the key tenets of the chatbot code of ethics, and it is essential to maintaining a good relationship with your users and customers.

How can you get started with building a chatbot for your business? Learn more about getting started with Watson Conversation Service.

 

Learn to build a chatbot with our free 30-day Bluemix trial.

Client Engagement Leader, IBM Watson and Cloud Platform

More Conversational Services stories
September 10, 2018

Clearscope makes it easy to improve organic search with Watson AI

Clearscope is able to use IBM Watson to reverse engineer the search engine process so content marketers can understand the key concepts surrounding the topics they want to write about, without having to duplicate content or resort to keyword stuffing.

Continue reading

September 7, 2018

IBM named a “Leader” in Forrester Predictive Analytics and Machine Learning report

Watson Studio was named a “Leader” in the Forrester Q3 2018 Wave report on Multimodal Predictive Analytics And Machine Learning Solutions. According to the Forrester report, IBM Watson Studio is a “perfectly balanced PAML solution for enterprise data science teams.

Continue reading

August 29, 2018

Meet the faces behind conversational AI

Who are the faces behind successful AI assistants? Meet one of them: creative writer for conversational AI Maya Hausammann, who crafts responses for Autodesk’s Virtual Assistant, AVA.

Continue reading