Design thinking for chatbots

4 min read
Will Fanguy
  •  Mar 5, 2018
Link copied to clipboard

Chatbots, artificial intelligence, and voice-driven assistants are all rising in popularity and prevalence as the new, cool technologies that are going to lead us into the future. But at what cost? If we’re spending more time interacting with robots than with humans, should we take the time to apply design thinking in order to give these interfaces a more human touch?

By applying the key tenants of design thinking to our conversational technology design process, we reveal opportunities to help these interfaces be more user-centered. Instead of making the most effective and efficient bot possible, we design moments of surprise and delight that keep our users coming back.

Let’s take a look at the opportunities offered by designing a chatbot (one of the more approachable conversational techniques) and the ways you can use design thinking to create a great experience for your users that also makes good on business goals along the way.

First things first: Needs vs. wants

The first step in designing a great chatbot is asking yourself (and by extension, your users), “Do we even need a chatbot?

Why are you building a chatbot? What outcome are you trying to drive for your users? The “why” of designing should always come before the “how.” If you’re trying to solve more technical support problems in less time, you can build a new chatbot or you can hire more support folks. Factors like cost and efficiency play a role in this decision.

Related: The ultimate guide to chatbots

Designer Yogesh Moorjani suggests using the “So What?” technique to determine if your plan for a chatbot is the most feasible way to invest your resources. Imagine sharing your chatbot idea with a very skeptical friend, who always responds with “So what?” Use this opportunity before you’ve made a large time investment to examine the practicality of your plan.

“The ‘why’ of designing should always come before the ‘how.'”

Twitter Logo

Using design thinking to create opportunities for delight and surprise

If you’ve made it this far, you’ve come to the conclusion that designing a chatbot is going to solve problems for both you and your users. That’s great, but the hard work is just getting underway.

By utilizing the stages of the design thinking process during the course of designing your bot, you’ll find unique opportunities to delight and surprise your users while also focusing on a user-driven experience.

For those new to the process, we’ll be looking at design thinking as the five-stage model proposed by the Hasso-Plattner Institute of Design at Stanford ( The five stages of design thinking are: Empathize, Define, Ideate, Prototype, and Test.


You’ve already started the first step in using design thinking in your chatbot design. By examining the “why” behind your chatbot, you’ve started thinking from the mindset of your users.

To empathize, you need to do your research. Consult with experts (even if that’s your customer support team), talk to your users, and immerse yourself in the environment. If you plan to make a Facebook Messenger bot (to add to the more than 100k already available), spend time speaking with your users via the app. Find out what their needs are. Seek out pain points. Set aside your own assumptions and learn. This is a perfect example of the old adage “An ounce of prevention is worth a pound of cure.”


Time to define the problem. You’ve got business goals you want to achieve, and you’ve got mounds of research from the time you spent empathizing with your users.

Your goal here is to define your problem in a human-centered (not business-centered) way. Don’t frame it as “We need to build a chatbot to lower our response time for support tickets.” Look at it more like “People don’t want to spend a bunch of time looking for a solution or a help doc.” This approach presents a larger, more open-ended problem, and that’ll help out when you get into the next step, Ideate.


Now comes the fun part. You get to come up with ideas, throw them at the digital wall, and see what sticks. You know your users, and you’ve got a well-defined, human-centered problem. Let’s ideate.

Your goal here isn’t to come up with The One Best Idea™. It’s to generate as many ideas as possible. Think freely (or “outside the box,” if you must). There are plenty of approaches to this part of the process like brainstorming, mind mapping, SCAMPER, and my personal favorite, Worst Possible Idea:

Worst Possible Idea is a highly effective method that you can use to get the creative juices flowing and help those who are not so confident in expressing themselves by flipping the brainstorm on its head. It’s a lot of fun too. Instead of going for good ideas and putting the pressure on, call for the worst possible ideas your team can come up with. Doing this relieves any anxiety and self-confidence issues and allows people to be more playful and adventurous, as they know their ideas are most certainly not going to be scrutinized for missing the mark. It's way easier to say, ‘hey, no that's not bad enough’ than the opposite.

In a field as new and unexplored as chatbot design, this stage of design thinking could lead you almost anywhere. Good news: That’s okay. This isn’t the time to get stuck on voice and tone or types of prompts. This is the time when you start to think about human vs. automated responses, where the answers are going to come from, and how deep you want to go in an effort to save your users time and potential frustration.


The next step is to take some of those ideas and make them real. This is an experimental phase, and the goal is to identify the best possible solutions for the problems identified during the first three stages.

Your design team will produce some low-fi, scaled-down versions of your chatbot (and its guiding logic) in an effort to find what works. Every idea that survived the transition into Prototyping will either be rejected (which is what will happen to most of them) or accepted, revised, and improved.

“Set aside your own assumptions and learn.”

Twitter Logo

By the end of this part of the process, your design team will have a better idea of the constraints and problems that come with designing and implementing a chatbot, and they’ll have a more informed perspective of how real users would behave, think, and feel when interacting with your new virtual friend.

Again, because this is a uncharted field, your prototypes might be a little rough around the edges. You’re exploring new ground here. Don’t be afraid to take a chance on a new idea or to toss it all out and start again.


This is where all the hard work pays off. The Testing stage is where your designers, your researchers, and possibly even some of your users come together to test the more polished prototypes that were the results of your prototyping.

While this is the fifth step that’s been outlined, it doesn’t necessarily have to be the last. As a matter of fact, most of the time the information and feedback we gather in the Test stage leads us to re-define our problem or to better empathize with our users.

Design thinking is non-linear.Twitter Logo Maybe you’re conducting two steps at the same time, or maybe your designers are prototyping as they ideate in order to better visualize the solutions.

In the case of a chatbot, you might learn that while you wanted to develop a solution for your users to get faster answers to help questions, what they’d really benefit from is a more structured onboarding process.

That’s great! Take what you’ve learned, re-frame the problem in a user-centered way, and head back to Ideate. As long as you’re making it about the users, you’re free to go in whatever direction the design thinking process takes you.

Collaborate in real time on a digital whiteboard