I have committed to build up a range of experience prototypes throughout my major project and this week I began the first of these prototypes. After a great interview with Michael Owen Liston he spoke about how he used TextIt for his major project with refugees in Denmark. I won’t do his project any justice in this quick blog post so I would encourage you to read more about it here
Eating the Elephant
I hate business clichés as much as the next person but I learnt a valuable lesson this week. I knew what I wanted to test with this first prototype – The assumption that people instinctively know how to chat with a bot and also know what a bot is.
However I got carried away and tried to design the context in which this would make sense. Would I be a business? Would I be an AI assistant? Would I be a service? Would I work in a group chat? Would it be one to one interaction? I got bogged down in the details and technical challenge each of these questions brought about and lost a couple of days trying to answer these. Number one issue here being my assumption that I could pick up a new tool like TextIt and just get working with it. I was at this point beginning to lose time and focus on my initial question.
Another lesson learnt is to not under estimate the time required to pick up a new tool like Textit and start working with it quickly. I will say that the service is easy to use, straightforward to get a test up and running and has an immediate impact on seeing it’s application. Where it gets tricky is twofold – One to figure out the integration with Facebook Messenger and second is to design a scenario that makes sense to the user.
I won’t bore you with the details of how to integrate it successfully with Facebook messenger. More so as there is a pretty decent post here about to do it. The one section it took me time to figure out was that for each user you wish to test with you need to set them up as a ‘Tester’ in your Facebook developer portal. This is assuming you have not had the app published for public consumption by Facebook. This requires submission for approval by Facebook and in the context of my fake business it didn’t make sense to do.
Questions to answer
Textit is a service to allow you to build complex flow’s which a user can respond to through multiple channels – SMS, FB, Telegram etc
In order to start answering my initial assumptions I designed a flow which would question a user on their previous knowledge or experience with chat bots. The flow is built up in a series of questions that the user answers with ‘YES’, ‘NO’ or free text responses.
Release the bots!
I sent out the this initial prototype to 10 users for testing. The feedback and engagement was overwhelmingly positive. With only one failed test (the user was unable to access the bot due to geo-location reasons) I had some very quick and early insights.
In answering the question – How would you feel about interacting with a chat bot, I had several responses along these lines:
“I would rather talk to a human. I feel like they could answer my question better…”
There was a clear hesitation about the quality of response one would get with talking to a bot.
However many users were equally impressed with the convenience, speed and supposed accuracy of the chat bot.
In the post experiment interviews conducted I was able to gather some really valuable insights. People were about to imagine the context in which this UI would be successful;
Speed of access to a response is the biggest advantage. Most phone calls for example you have to listen, press this, that and the other key, wait 10 mins to be connected to someone etc. I find that really hard with (a) how busy I am and (b) kids killing each other in the background. With a text type service I can deal with it whilst I’m doing another task and it doesn’t matter where I am. If it’s a really noisy place I don’t have to worry about back ground noise interrupting the call and if it’s a really quiet place (cinema) then I don’t have to worry about me being quiet.
You can view the prototype below