The Guide To Designing A Magical Chatbot Experience — Part 2

Phase

Bots should be aware of contextual and conversation history, so they need to be able to keep track of users in different phases of their interaction.

Phases are major shifts in the mindset of the user, usually over longer periods of time. This is not referring to the different steps of a shorter conversation.

A traveller who has booked a flight is now ready to book a hotel. Someone looking for a vacation destination is definitely not ready to look at hotels.

Of the different personas and their intentions, what phase are they in with their intention? Are they pre-action, mid-action, post-action?

Travel bot example phases:

Travel assistant bot phases

Each category consists of a very different user mindset, and the user will be looking to accomplish very different things.

Consider following up at the end of their natural lifecycle. You already know what hotel they booked and when, why not check in to see how their stay was? Provide better recommendations in the future based on that feedback.

Notice that in this example, the phases apply to all the user personas, however the information and actions they seek for each phase may be different. For your bot, the phases may only apply to certain users.

Contexts

Contexts are the information you need to help place the user in the correct user story. Using context, you can infer their persona, intention, and phase.

Sometimes the context is available from their profile or channel and the fact that they are talking to your bot is a context. More often, you need to gather context from their queries, or from their conversation history. Worst case, you will need to actually ask them to provide some context for you.

Guidelines:

  • Infer context as much as possible.
  • Favor asking for information at the point that it is needed, not before.
  • Do not overwhelm the user with a bunch of questions during on boarding.
  • Be creative. Instead of asking them a bunch of questions outright, try to think about what you can infer from fewer questions.

Based on your answers from the Persona, Intention and Phase section, think of the different contexts required to put the user in those buckets. You will want to keep the number of contexts minimal at first, but you should expand and give them more depth as your bot grows.

Travel bot example contexts:

Travel assistant contexts

If you have a user’s Nationality already, getting their current location can help you figure out a lot about their Persona & Phase. If they are in their home country, they are probably not a digital nomad or expat. If they are already located in the country they wish to travel to, they are in the arrived Phase.

Notice that Persona, Intention, and Phase are contexts, but we often use other contexts to infer those contexts.

read original article at https://medium.com/@_TomHoward/the-guide-to-designing-a-magical-chatbot-experience-part-2-f76be35fc68e?source=rss——artificial_intelligence-5