Conversational Design – A date with Watson Conversational chatbot
I don’t know about you guys…but sometimes I feel when I have conversations with folks, I tend to think how the hell did I get into this as either the conversation has got into something I didn’t intend to or the other person has no clue of what Iam talking to.LOL! When I feel that way with “full” humans, Iam kind of skeptical about any conversations with man-made chat bots. This is where Conversational Chatbot comes in.
Conversation meaning is “the informal exchange of ideas by spoken words” per google search. The best synonym I like for conversation is “heart-to-heart” conversations. Who doesn’t love a heart to heart connection and conversation?
So, I decided to try some “heart-to-heart” conversations with some chatbots out there to see how effective they have designed it.
Here is my first part and my first try was with Watson Assistant Chat App which is trained for specific set of car capabilities.
Available at https://watson-assistant-demo.ng.bluemix.net
This is how the conversation went…
I like the fact that it cared to ask what cuisine Iam interested in. However, when I didn’t give any choice and asked to pick one for me, the bot struggled. Also, when I said Iam not in a mood for tacos, it made it worse as it lacks sentiment analysis. More trouble started when it came down to when I asked to take me to the closest location as still it was asking whether the first, third or nearest and so on. I would have thought closest is synonymous to nearest and would have given me the nearest location.
From a Conversational design evaluation perspective, Iam looking at the following attributes for my rating –
- Personality – How does this bot differ from others? Does it have a taste of humor in its response?
- Content – Every word uttered by the bot matters, and use of emojis, gifs, images, videos, makes it more presentable
- Flow – A designer must create and build all of the paths the user could take to reach the end goal. The more storyline , path you have the better the experience in terms of the chat bot experience.
- Strategy – This involves the effort put in building the main conversational flows, determining the feature set including any web hooks/integrations we should have, text vs speech input mode, etc.,
Based on the above criteria, I gave the above bot created using Watson a rating of 5 out of 10 as it lacks personality, and fun content. The flow is moderately well designed but however there is not much support of strategy around it as it’s a simple example. Hope this gives an idea of how conversational design shall be applied to chatbots!
Let’s converse to build a conversational chatbot!