Skip to main content

Ending conversations

As with any human to human interaction, there’s always an ending to give participants closure and end on a positive note. This is the same for every chatbot interaction, including industrial chatbots. Although it seems an easy communicative function, ending interactions well can achieve multiple goals that help users, chatbot teams and chatbots. It’s an opportunity to get useful feedback to train your chatbot, create a clear signal that the interaction is complete, and set the tone for the next interaction interactions.

Examples​

Ending conversations: Short example

This is a simple, effective end showing the supportive nature of the chatbot.

Ending conversations: Longer example

This is a more empathetic ending, highlighting the successful support given and the support it will continue to provide the user.

Ending conversations: Example with apology

Here the chatbot offers an apology for not being able to support the user, explains why, and then closes with a more positive note to try to compensate for its failure. While this is a negative end to the interaction, providing a clear reason for its limitations and then offering to make amends with the next query, helps soften the disappointment.

Dos and Don’ts​

  • Do consider adding a time-out feature
  • Do consider adding thumbs up and thumbs down icons to get feedback
  • Don’t loop back and have the chatbot reintroduce themselves