Designing for Bot
Conversation, but not as you know it…
As a teacher, a writer, and a lover of all things techie, I love to learn. Designing conversations for chatbots then turns out to be the dream role. It very rapidly became apparent that there was so much I didn’t know and needed to learn, quickly.
A quick google on how to design a chatbot conversation will bring lots of really useful and not so useful articles on where to start. Many come from a developer perspective with a deep understanding of the platforms on which they build the bots. Not ideal for a newbie non dev.
Chatbots Magazine is great and there are a few good short courses. UX for Voice on LinkedIn and Voice User Experience Design on Udemy were helpful.
One of the first things many writers suggest is designing a persona for your chatbot. Your persona should be a fully rounded character and as well as being good fun to create, there is evidence to suggest that chatbots with personality are more engaging and encourage continued interaction. It also helps with understanding your user. If you want to reflect who the user is, you need to understand their context and their needs.
Many articles suggest starting with the Myers Briggs personality chart. Not a bad tool, but the persona needs to be in line with corporate image, while still being relatable to the user. After a slight deviation into the realms of UX fundamentals, emotional intelligence and affinity theory, a character emerged. Our persona is always informal (this is conversation), slightly extrovert with a sense of humour (easily relatable), encouraging and efficient (in line with the corporate), and always agreeable and openminded (the bot is there to serve after all).
Whilst I had spent time reading the books and guidelines on design, Emotional Design by Donald A. Norman is very enjoyable, it’s at this point I should have taken time to read the testing documents. It would have saved so much time in not having to re-write text later when you are submitting your bot.
Many of the articles I read, suggested starting conversation design in the shortest, easiest way: the happy path. This proved very effective, write the conversations first, filling them in, scrawling them down on paper along with all the ways you can think of that someone might ask the question. Write down what might happen, how the bot might reply, with variety. Would the user follow on with other questions, does the bot need to remember anything? Put down everything you can think of with how the conversation might go, but start with the most straight forward first. Only then start on a diagram.
I’ve also learned how super, uber, crazy important it is to talk to, learn from, ask questions and work with the rest of the team. I know it sounds obvious but it’s no good saying, ‘the bot should say this’ or ‘the customers want this’ when it’s impossible to build that way or someone can suggest something even better.
The team I work in is full of intelligent, creative, helpful people who do all sorts of things and so have all sorts of different strengths, ideas, experience and solutions to all kinds of problems. What a resource, I hope your team is just as fab.
The last six months have proved to be an incredibly steep learning curve but there are a lot of resources out there, and here in the office too. Make use of all of them and enjoy the ride as the sense of satisfaction when your conversation comes to life is definitely worth it.
