Conversational Interfaces -To Do or Not To Do

For quite some time, software product designers have been dealing with only couple of user interfaces — Web and Mobile.

Now with the advent of AI and Machine Learning they have a new beast to slay — Conversational Interface, commonly implemented in the form of a Chatbot.

If you own a growing or successful business, you may already be thinking to jump the bandwagon. As a founder of CricHeroes, I am also contemplating to provide conversational interface to my users.

But here is the thing, How do you decide?

Based on my experience and understanding of software product design so far, thinking through following questions may help you decide.

Is it faster than using your website or app normally?

In most cases, a well-thought navigation is all you need for your website or app. Typing is work. If I am going to find what I need faster by just navigating your site (or app) on my own, I am not going to use your chatbot.

Here is what Tapzo Founder, Ankur Singla, learned the hard way.

People often confuse the debate with bots, AI etc. but does it really make sense to chat to get a cab when you can simply tap once and book (no matter what you think, our user data was very obvious on this. The answer is no.).

Does it add more value?

Can your Chatbot give me something more, which I will otherwise not find quickly by browsing? For e.g. an Educational Chatbot giving personalised tips to me as a parent based on my child’s progress. If yes, go for it.

Are your users willing to type?

Typing is not for everyone. You have to think, frame a question and know basic English to type a question. Not everyone is willing to do that. So know your target audience before making a call.

Are you willing to handle edge cases?

No matter how many rules you add or train your AI engine, you will miss out some edge cases. Your chatbot should not be giving the default answer, “Sorry, I didn’t get that.”, too many times. It is simply a bad user experience.

Will it be rule-based or machine-learning based?

Rule-based chatbot is easy to make but you won’t be able to cover everything. When you give your users a free hand, they are going to type anything. Are you ok with that?

Chatbot based on Machine-learning engines are supposedly smarter. They can cover up on their own if something unexpected is asked and become better at it next time. Are you willing to spend time and money building that?

Will it live inside your app or on third-party platform?

Many of the chatbots today are available on third-party platforms like Facebook Messenger. Is your target audience on FB messenger? If yes, are they willing to open FB messenger and interact with your chatbot there?

Do you want to control the experience?

You will not be able to control the entire experience with Chatbots. Users may type gibberish and still expect correct answers. When they are using an app or website with traditional navigation and interface, the entire experience is much more controlled. You exactly know what will happen if user clicks anything. If being in control is a must for you, you should not go for conversational interface.

So, that’s it? May be I have missed out a few points but you get the idea.