You’ve taken the first step by creating your chatbot persona. Now it’s time for your first sample dialogue: a practice conversation between your chatbot and your clients.
Sample dialogue is the foundation of all your bot's conversations. When writing dialogues for a chatbot or voice assistant, you write conversationally. In other words: how people talk. And how do you find out how people talk? By doing it yourself! That way, you discover the natural flow of a conversation, how to provide information, and which strategies you can use in case if a conversation gets stuck.
Need some help in creating your first sample dialogue?
Before we start our sample dialogue, we need to get to know our actors: the chatbot and the user. We do this by filling out a worksheet. If you've set up the chatbot persona beforehand, we already know the bot's personality.
The second actor is a user. We decide who speaks and under which circumstances. Have they used a chatbot before? Are they in a hurry? The more we know about our user, the easier it is to play their role.
Next, we choose a topic: a use case. What does our user want to achieve? Are they looking for opening times? Do they want to return an order?
In the previous step you’ve determined who talks with our chatbot and why. Time to begin! One person plays the chatbot, the second plays the user. The two actors sit back to back. This is important, because we shouldn’t be distracted by body language. A chatbot uses only words to communicate.
While the actors practice their conversation, the others take notes. They pay attention to the conversation’s structure: how it starts, how it ends and the strategies the actors use. What does the bot say when the client doesn’t answer? And what does the client say when they are confused?
Once the conversation runs flawlessly, it’s time to write it out. We start with the happy flow: the scenario where the conversation runs smoothly from start to finish. We make sure the chatbot’s personality shines through every sentence. After that, we write content for the unhappy flows - if something goes wrong. For example: the bot is unable to do something or doesn’t understand the user’s answer.
We thoroughly test our content before publishing. We do this by using Wizard of Oz testing. It works as follows: one person plays the bot, but they can only use the written content when speaking. Others play the users and ask the questions. If an answer is missing, we will find out immediately and create it. This means you'll find the missing gaps before publishing, saving you time in the long run.
If your bot talks, like Alexa or Siri, it’s even more important that the conversation feels ‘human’. You want your customer to talk to your bot naturally, just like the way they would talk to a person.
Our conversation designers guide you every step of the way. Whether it’s your first sample dialogue or the hundredth, we make the preparations and ask the right questions. Together we build the best foundation for every conversation.