In order to work with Answer Bot effectively, it’s important for agents to understand Answer Bot’s intended purpose, how they can use it to help their customers, and what they can do to improve Answer Bot’s performance.
This article includes the following topics:
What is Answer Bot?
Answer Bot uses machine learning to measure the similarity between knowledge base articles and tickets, and uses that information to respond to support tickets with an automated email that lists relevant articles. This encourages self-service in your customer base, and raises awareness of your knowledge base offerings.
As more customers are able to solve their simpler problems through self-service, agent time is freed up to tackle more complex problems.
Using Answer Bot
Answer Bot results, and the end-user's response to them, appear in the comment stream of a support request, but it does not require any agent interaction.
However, agents can glean important information from Answer Bot results, and use that information to improve their customer service.
When the Answer Bot trigger fires, suggested articles are sent to the requester. These articles appear in the comment stream of the generated support request:
When the end-user interacts with any of the articles, the Answer Bot comment in the support request is updated:
- indicates the end-user has clicked on the link, and viewed the suggested article.
- indicates the end-user has clicked on the link, viewed the article, and responded that the article did not answer their question.
When an end-user views an article and finds it answers their question, they can respond by clicking Yes, close my request. This adds an internal note to the support request, and marks it as Solved:
Improving Answer Bot performance
Agents can assess the value of the answers given, and improve the algorithm from the comment stream.
To improve the Answer Bot algorithm
- In the ticket's comment stream, click Improve answers to open the article selector:
- Click the checkboxes to mark any off-topic articles.
- Click Done. Removed items still appear in the list of suggested articles, but the title appears with strikethrough font: