Published in


Chatbot Design: What Makes a Good UX

AI like those we see in movies is not practical!

When we think of the word ‘AI’, chances are it conjures up pop culture depictions like Iron Man’s JARVIS, HAL 9000 from A Space Odyssey, or GLaDOS in our heads.

What do these depictions have in common? The typical AI-gone-wrong trope aside, they are all rather chatty AI. (After all, they are their own characters and villains, so it makes sense for them to have quite a few lines of dialogue.) But, in real life, we rarely encounter such talkative, somewhat human-like AI. Why is that?

Do you really need a human-like AI?

There are some fundamental difficulties in developing AI that can converse exactly like a human. But besides that, it’s expensive to do so — in most cases, it simply wouldn’t make commercial sense. And would it actually be useful, in the first place? If you think about it,

does Tony Stark really need JARVIS to do more than execute his commands?

This is a good starting point when thinking about designing chatbots for commercial use. The AI that chatbots are going to rely on is probably not sophisticated enough to fully mimic human conversation anyway. Chatbots are supposed to be better than humans when it comes to specific tasks. Although humans can understand complex requests easily, chatbots can crawl through thousands of information from the knowledge base much faster. It would make more practical sense to build a chatbot-human experience that ensures a chatbot does its job of retrieving data quickly than trying to mimic human interaction which is not 100% possible.

Now, after setting aside romantic ideas of creating a Turing Test candidate, most importantly, we need to consider what your design goals actually are. A good user experience (UX) process can help with that.

Theory into practice — building conversational UX

A summary of the UX process

1. Identifying Design Goals

First, of course, we need to consider the business goals of the company or the client. Why are we being asked to create a chatbot? What is the chatbot supposed to do?

2. User Research

For a good design, we also need to consider the end users’ experience. How would the chatbot be useful for them? What problems can it help them with? What needs does it address?

This is where we need to do some research on the needs of our end users. Ideally, this should be done by talking to them directly through one-on-one interviews or even focus groups. By asking carefully crafted questions, we can better understand what they really need and what their relevant habits are like without making any assumptions. For example, what kind of questions do they usually ask this type of chatbot? What do they do when they are unable to get the answers they need easily?

3. Creating Personas and Identifying the Problem

From here, we can create a few personas that encapsulate the profiles and needs of our end-users, and a problem statement or two to summarise what exactly the chatbot is supposed to help with.

4. Solutioning

With these to guide our subsequent efforts, we can then figure out the solution in a coherent way — what should the chatbot do? For example, should it present users with interactive menus they can easily navigate to get what they need? Or should it invite users to ask open-ended questions and then respond with relevant answers?

Focus on meeting the business goals and fulfilling user needs. Don’t go out of your way to make the bot sound as human as possible if it means compromising on the bot’s effectiveness and ease of use.

Now, let’s talk about conversations.

We are starting to get into conversation design here. There are some basic principles we should keep in mind to ensure good conversational UX.

A well-established set of guidelines came from Paul Grice, a scholar of language, who described four maxims for cooperative communication that make conversations more effective.

Grice’s maxims for effective communication

These maxims help us keep the conversations focused on user needs and reduce user frustration when chatting with the bot designed, no matter what the chatbot is for. And happier conversations will definitely contribute to higher user satisfaction.


Chatbots may not be able to engage in witty repartee but, with an in-depth UX process and by incorporating effective conversation design principles, they can at least do their jobs well and create good user experiences. Now imagine how much more efficient AI villains would be if they spent less time talking and more time executing their plans. That’s a scary thought!

With the ease to build a customizable chatbot and the rise in messaging apps, this proves the growth and strength that will push the trend of chatbots forward.

2359’s in-depth knowledge of Chatbot solutions and innovation strategies has helped our clients with handling the influx of customer requests as well as delivering business values by streamlining internal workflows.

Interested to find out more? We’d love to have a chat with you!



Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store