How to design for a better conversational user experience
Quick pointers on getting to start designing better conversational user experience in the realms of chat bots and intelligent assistants.
Define your company or product’s bot personality — This step goes a long way in the start of the relationship. It could be a bot that reflects your brand, it could be a micro brand. Spend time in defining the goals for your bot’s personality, this is not just a marketing task, so take ownership and collaborate across your team to define this. Once the personality is defined, share it with your entire team. Make the bot a part of your team by making swag like posters, t-shirts, coffee cups etc. This should help the entire team focus on how they want to build the bot for your end users.
Minimum dimensions to focus on for your bot are listed as follows
- Tone or mood
Explore interaction solutions beyond chat sessions — With the advent of so many platforms opening up, it is easy to get carried away. Try not to focus on the hype and restrict your MVP’s based on FB Messenger etc. Though it is a very effective way to start there, do not forget to explore other touch points. Example — Should my bot assist during checkout? How should I make the bot surface in my traditional channels like websites, native apps on phones, tablets and desktops?
- Do not constrain your bot to just chat windows
- Create bot presence in other user interface channels or patterns
Create a very basic set of UI components for each interaction channels — The moment you start talking about NLP, everyone starts thinking about the user “talking” to the bot. This is rarely the case. If you research or spend time observing people around you, not many people “talk”, they type. The utopian era where users would loose their inhibitions to talk openly to a computer in our society is still far away. Since AI enabled bots will be a new phenomenon for most users in 2016 and in the coming few years, we are actually designing for early adopters. So start with a very basic set of UI controls that can be reused across chat channels and other traditional web UI or native OS UI.
- Design cards or tile views that would accommodate text, image and video
- Make it scalable, graceful and crisp on chat channels, dashboards and catalog views on form factors of a phone, tablet and browsers on desktops. If you have the bandwidth and the inclination, give VR a go and let us know what you learned from it. It is not easy, that much is clear from the Google’s VR team.
- Design CTA buttons and keep them as singular action, at the max use segmented button controls for more actions. Do not use too many actions, you need to make the bot come up with very narrow choices for the user, otherwise, the bot is going to be perceived as just another extension of a dumb computer.
Define your user interaction flows for handing off to existing solutions and taking users back from that channel to chat conversations. Example — What happens when user taps or clicks on the CTA? How do I take them back to a conversation once I have made them complete a task or transaction in other interaction points?
Have ways to measure how your bot is performing and make that information available to your team. This could be a completely new area but how do you know your bot is meeting your goals. Is it actually improving your product or brands UX goals or is it hindering it? This is a very crucial bit for improving conversational user experiences. Examples — Are there more chat sessions with the bots. Can we sense frustrations from the users side? Is the bot receiving kudos or thank you messages. Is the bot driving CTAs? Think about innovative ways to squeeze in success measuring criterions into the bot.
Bots and NLP are just riding the wave again, it is interesting to highlight that such interaction channels were already there for the last 20 plus years or so, it just did not get the success it deserved due to the market and times not being ready for it.