Design process

updated on:

6 Mar



Conversational UI: How to Create а Brisk Human-Machine Dialogue


min to read

Natalia Borysko

Writer at Eleken

Natalia Borysko

Writer at Eleken

Table of contents

In my childhood, impressed by Star Wars, I dreamt about having robot friends like R2-D2 and C-3PO. Could I ever have thought that we would become a bit of Luke Skywalker chatting with machines in less than two decades? 

Yet not so smart and empathetic, chatbots help businesses boost customer engagement and increase work efficiency through close-to-natural communication with users. On the other hand, it turns into quite a frustrating experience when a conversation with a chatbot hits a dead-end. 

To create an effective conversational interface for customers, Eleken designers meticulously searched answers to critical questions. How can we make bots act more humanlike? What differentiates a quality chatbot from those that spur customers to abandon a dialogue? What challenges do chatbot developers face and how to create a conversational user interface that will benefit your brand? 

If you feel like this is what you need to know, keep reading this post. 

And may the Force be with you ©

What is conversational UI?

Simply put, it’s an interface connecting a user and a digital product by text or voice. Conversational UI translates human language to a computer and other way round. This became possible due to the rise of artificial intelligence and NLP (natural language processing) technology in particular.

Conversational interfaces are extremely important in the customer service realm, where agents should always be ready to accept and process clients’ inquiries. During peak or non-working hours, when customer support isn’t up and running, chatbots can address some customers’ questions and route the communication further to a human “colleague”.

In all fairness, it has to be added, a customer experience depends much on chatbot communication abilities. Contextual AI-driven and rule-based bots are more flexible in understanding and interpreting users’ queries than chatbots with preestablished answers that narrow communication to limited algorithms. 

And here you might be furrowing your brows in bewilderment: “Wait, what are you talking about? Too many cryptic words in one sentence.” 

I anticipate this, so here the explanation goes.

Types of conversational UI

If we divide conversational interfaces into two groups, there would be chatbots and voice assistants. Even though we concentrate on chatbots in this article, voice assistants shouldn’t go unmentioned.

Siri by Apple, Microsoft’s Cortana, and Google Assistant use voice recognition and natural language processing to understand a human’s commands and give a relevant answer. The AI technologies voice assistants are based on are complex and costly. Thus, for the time being, only tech giants can afford to invest in voice bots development.

Unlike their voice counterparts, chatbots became quite a widespread solution online businesses adopt to enhance their interaction with customers. 

Depending on core technology, chatbots fall into two categories:

  • AI-driven
  • Rule-based

Chatbots powered by artificial intelligence, namely natural language processing and machine learning, can literally read between the lines. They not only understand users’ queries but also give relevant responses based on the context analysis. 

What is thrilling about AI-driven chatbots that they:

  • Absorb immense amount of data and continuously learn
  • Understand different languages (at least those with Latin alphabet should be no problem)
  • Comprehend users behavioral patterns
  • And (that’s mind-blowing!) they can make decisions

Rule-based chatbots yield to their AI relatives. A set of rules predetermines their interaction with customers and gives no space for improvisation. However, this type of bots is less expensive and easier to integrate into the various systems. The more detailed algorithm a chatbot has on the backend, the better the communication experience a user ultimately receives.

rule-based chatbot example
A chatbot-humans conversation mapping

Regardless of the chatbots’ usefulness for business, there’s a catch. People don’t like chatbots...yet.

Conversational UI challenge

Before coming to this conclusion, I decided to conduct my personal tiny UX research. I surveyed only one but reliable participant whose opinion I deeply respect - my husband (he’s a software engineer and a tech geek). 

My questions were whether he had ever used a chatbot and what his experience was. It turned out that my respondent tries avoiding communication with chatbots because of two reasons (have you noticed this “2” number is chasing me throughout the article? Let it be the “number of the day” then.

So, the reasons are:

  • It costs extra effort explaining the issue in the way a chatbot could understand
  • The bot failed to comprehend the issue, so it could neither help nor connect to a human agent

Waste of time and nerves. 

Well, not inspiring results. But what do other people think? So I googled and found the research carried out by Userlike guys that proved my concerns.

More than 50% of the surveyed audience was disappointed with the chatbot’s incapability to solve the issue. Around 40% of respondents claimed the bot couldn’t understand the problem. 

disadvantages of chatbots
The most impactful negative aspects of communication with a chatbot

However, 70% admitted that the chatbot answered them quickly, and 40% mentioned the chatbot could assist them outside of regular working hours.

advantages of interacting with the chatbot
The positive experience of interacting with the chatbot

According to the following graph, people would like to use chatbots rather as a link between them and a human agent than a full-fledged assistant. 

do people love chatbots? survey

The conversational user interface design needs to generate the best customer experience possible to show users the huge chatbot’s potential. Every detail in conversational UI/UX should be considered to mitigate the skepticism of those customers whose initial experience was corrupted by a low-quality chatbot.

On the other hand, a well-designed bot will help you:

  • Strengthen brand image
  • Assist with handling multiple tasks
  • Reduce service costs

The biggest challenge is making chatbots more human-like without pretending to be real humans (as this deceit can provoke even more negative emotions).

9 principles to humanize chatbots UI

Here are some principles to help you create chatbots your customers would love to talk to.

  1. Make a chatbot more specific

To avoid customers’ judgment that your chatbot is incapable of helping them, be more specific in what your chatbot can offer to customers. If a bot can accomplish simple, unambiguous tasks like help customers place an order, check order status, or choose food from a menu, that would be helpful. In case you aren’t sure your chatbot is trained enough to handle complex requests, think of limiting the options it can help with. 

The chatbot on the image below asks customers what they’re craving without options’ limitation, therefore can’t eventually understand the responses. 

communication with a chatbot example

  1.  Use multiple-choice buttons

AI-driven bots learn to recognize and understand human language common patterns thanks to NLP technology. However, the problems happen when people alter their natural language in the heat of aspiration to help bots better understand them. And this is a total fiasco. 

The conversational interface designed to facilitate the interaction with customers leads to a conversation dead-end. Persuasive design techniques can help rectify this problem. For example, several options of answers, realized in the interface by multi-choice buttons, limit a user to a range of offered selections.

multiple-choice buttons for a chatbot
Multi-choice buttons ease the communication between a user and a chatbot
  1. “Give me more emotions!”

Emotions are an invisible glue that sticks us to screens when watching a heartbreaking drama. In messaging, we use emoticons, images, and gifs to convey our emotions and make a text less dry and soulless. The same approach will work for conversational interface design as well. 

chatbot window example
Emoticons’ injection creates a more friendly communication vibe
  1. Give feedback, react, reward

It’s crucial for the chatbot to identify peak moments in dialogue and adequately react - encourage, congratulate, or cheer the client up. I loved this natural dialog between the Freshchat bot by Freshdesk and a user. It sounds like a small chat between two schoolmates.

Freshdesk chatbot example
  1. Nudge customers in a right direction

Sometimes it’s necessary to give users a gentle push to perform a particular action. At the same time, a chatbot can reassure a customer that it’s okay to skip some action or come back later if they change their mind. It’s crucial for the user to have a feeling of a friend’s helping hand rather than a mentor’s instructions.

  1. Utilize selected words for commands

It’s great if you try enriching your chatbot’s vocabulary. However, using various words to mark the same functionality may lead customers to confusion. They simply won’t comprehend what actions they need to take if every time these actions are named differently. It would be better to be consistent and use some selected words throughout the conversation.

  1. Choose a chatbot’s voice and tone

The main thing here to remember is that a conversational interface should correlate with your brand values and act as a brand ambassador. The rest is up to you and your business to decide what voice your chatbot will have. It can even be like this one.

the voice and tone of a chatbot

  1. Create a character

Remember R2-D2? He wasn’t just a piece of metal and microchips. He had a personality. Your chatbot should also have. 

The most stunning example of a chatbot’s personality I’ve ever seen is an AI-driven bot Kuki (formerly known as Mitsuku). She’s extremely smart and ready to chat with you 24/7. 

Make sure your bot is a vivid character people can remember. 

the use of an artificial intelligence markup language by a chatbot to imitate human conversations
The bot uses an artificial intelligence markup language to imitate human conversations

However, it still should be a bot. Don’t try to delude customers that they’re talking to a real human. It may evoke a negative attitude to your brand when they reveal the deceit. And again, set your chatbot’s purpose first and think of a character afterward.

  1. Avoid a dead-end situation

Well, perhaps it’s not that easy task, but at least a chatbot must have a pre-established setting for the cases when it doesn’t know the answer. However, it shouldn’t be one phrase perpetually repeated. Also, it’s essential to offer a walkaround if the conversation hits a dead-end. The ultimate goal is to provide a customer with a great conversational user experience, so go from there.

A final word

Will chatbots reach the real human conversational level? I would vote for “yes”.

With artificial intelligence development, chatbots will become smarter and more capable of driving the conversation without embarrassing flubs. Our designers always keep a curious eye on the latest tech trends and are ready to apply the freshest knowledge in designing your chatbot. And here we have more about UI/UX trends and SaaS trends for 2021; read them on.