Design principles for Conversational Interface

Till the last decade, we thought of Automation as a futuristic technology. Well, the future is here!

Ever since the chatbots started making real impact on businesses around the globe, enterprises have incorporated them as a part of their business strategy and invested in them with devotion.

It is shifting our ideas about conventional Graphic User Interfaces, because chatbots essentially are All Talk and No Buttons.

What essentially are the guiding principles for conversational interfaces? Well, there are a lot! But I would list the indispensable 10 commandments here -

  1. Type Less — The entire idea is convenience, after all. Design conversations such that the users need not type a lot. It can be through efficient and specific routing of dialogs to suit user requirements along with added Quick Replies.

2. Informative responses — The responses should be engaging and informative whenever the intent is missing.It can be a call for the user to be more specific in queries or informing about an invalid action, the responses should always be crisp and clear.

3. Analyse Intent — What does the user want? A prerequisite for this would be a well defined goal for your bot. If the user is well informed about the knowledge and action domain of the bot, it will guide her to provide inputs that a bot is capable of handling. Once that is checked, it is important for the bot to clearly understand the intent of the conversation flow. For ex. For a survey bot, if I say — How to delete a survey and Delete a survey. The former is just a query (in which case it would show the necessary steps) while the latter is an action( survey would be deleted). Confusion between the two would be quite distressing for the end user.

4. Call-to-Actions — Despite all the magic, chatbots have limitations, of course. To keep the conversation in-flow, it is imperative to use Call-to-Action buttons or text which guides the user to the intended path of action. It reduces typing and hence is convenient for the user.

Check out for my next blog to know how you can increase engagement in your organization through data driven intelligent Call-to-Actions.

5. Undo Input — Save your users the pain of mincing over a spelling mistake or accidental inputs. The undo should be as easy as the regular GUIs. The liberty to err is indeed crucial from a user’s perspective.

6. Do not disturb — With chatbots you can allow notification control to users. Users can either set their preferences for some notifications or choose to not receive a particular type of prompt/notification. The bot, like humans should realise when their actions become uncalled-for.

7. Be in Context — Context is crucial for designing bots. It helps in reducing the friction of the conversation and helps the bot to grab the best out of user inputs.

A basic example would be that you should train your bot to recognise all possible ways of greeting someone like — Hi, Hello, Hey, etc. Thinking and framing these scenarios beforehand helps in enhancing the user experience.

8. Always measure — Data is the new oil, don’t forget to capture everything you can. Listen, note and never ask things you already know. For example, user’s name, employee id,etc. once added (either by user/organization) should be remembered for any future reference.

9. Create Delight and Engage Users — Always anticipate user’s response and lead the conversation in a way that he/she is also interested in heading right ahead with the bot. The conversation flow should always head towards the intended action for the user.

10. Robustness and Scalability — Robustness and scalability are desired of every product. Similarly, the bot should work as desired even under heavy load and unexpected user inputs. Also use cases you wish to cover can be scaled according to the customizations needed by the user.

Now, that you know all that you need to, ready to get things done?

If you like it faster and better, check out our Workplace bots for enhanced Employee Engagement and productivity at Kwench.