Foreseen and Unforeseen Interactions in Chatbots
Although there is a parallel between the traditional and the conversational UI, there are differences in the way of thinking about the interactions that the user may have during the experiment.
When the user starts a conversation with the robot / site / service, the interaction usually starts in an open field, where the user can type what he wants, this is the unforeseen way.
An example of this type of unplanned interaction is a customer service robot. “What can I help you with today?” Is an open-ended question that allows the user to type anything on the application, and depending on the answer, the robot needs to know if the user is having trouble connecting its internet or if its account for this month is delayed, two very different scenarios that can only be “understood” by the robot if there is a previous mapping of keywords.
But once the robot identifies the topic the user wants to talk to, a more guided moment begins in the conversation so it can identify how to do best to help.
The expected interaction is the one where you can know exactly what the user’s steps are. For example, I worked in a chatbot where the objective was to offer the user an estimate value of its electricity bill, asking questions such as: what type of meter, how many rooms do you have in your house, and in the end, the simulated value of the bill was displayed with the steps to migrate to that energy company.
The first challenge from UX’s point of view is knowing how to identify the best kind of interaction and path for each context. Knowing the right time to switch between foreseen and unforeseen events makes the conversation happen much more naturally and you get to the point where it will help the user in the task it is trying to do there.
Leandro Costa is a Brazilian creative developer based in Dublin and working as front-end developer & designer at Boys and Girls.
