Why ask your users for feedback?

Asking your users for feedback is a good way to assess the quality of the answers given by your chatbot. Depending on the kind of product or service you are providing, there are several ways to implement the user feedback functionality. These options can be different in their composition but also in their design.

The guide provided below will show you how to set up a simple feedback dialog for Conversational AI Cloud to use with Web Conversations.

Need more information about implementing feedback? View the developer documentation here.

Feedback Dialog

The following scenario will show you how to set up a simple feedback dialog for Conversational AI Cloud to use with Web Conversations.

This feedback dialog makes use of the following elements in Conversational AI Cloud:

• Metadata
• Dialogs
• Event

Metadata

First, you will need to create two metadata schemes:

feedbackScore with type “Simple list” and values -1 (negative) and 1 (positive)

Now create another metadata schema for showing the feedback:

Now that you have your metadata ready, you can start building your feedback dialog.

Feedback Dialog

Now we will create a simple feedback dialog flow. This feedback dialog will be shown to the end user when they have received an article answer or reach the end of a dialog flow. Our feedback dialog will look something like this:

Did this help to answer your question?

• Yes: Great! Thanks for your feedback.
• No: Sorry to hear that. Can you tell us in a few words how we can improve?
• End: Thank you for your valuable feedback. If you have any further questions, just ask!

On the positive node output, click on the “Context variables/Metadata” button, select the “Metadata” tab and add metadata “feedbackScore” with the value of 1 and “showFeedback” with false to the positive feedback node:

Next, add metadata to end node with the catchall:

Save the dialog.

💡Good to know: Adding the value “false” to this node will prevent the end user from getting stuck in a feedback loop. If you don’t add this value, they will be shown the feedback dialog again.

Event

Now we just need to create an event which will show the feedback dialog to the end user. Create an event called “show_feedback” and select the feedback dialog as an answer.

Web Conversations

The final step is setting up the feedback event in Web Conversations.
From the left menu, select “Behavior”

On the right side you can configure the behavior for the feedback. Use the following settings for the feedback event to show and select the feedback event that you just created in Conversational AI Cloud from the menu:

Be sure to save your configuration in Web Conversations and publish your project in Conversational AI Cloud. You're ready to test the feedback flow you've just created!

Note: If you are using V1 of Web Conversations, you can enable the feedback by selecting “UI Elements” and enabling the “Dialog feedback” option. The name of the event in Conversational AI Cloud must be “show_feedback” in order to work.

What Will You Do With the Feedback You Have Collected?

Choosing to add feedback functionality to your project will enable you to get more insight about what users think, and hereby draw a clear path on how to make the user experience better.

The Feedback Dashboard will help you analyze all feedback given by users in your project. You can look at the feedback you collected from different angles, for example, you can check how many answers received feedback over a certain period or you can choose to look at the distribution of positive and negative feedback over a number of answers. You can also decide to select a few answers and look closely at the comments that the users left for these.

Several reasons can lead a user to give negative feedback on an answer, such as the answer being unhelpful, incomplete or outdated. Once you have diagnosed the reason for negative feedback, there are a couple of things that you can do. You can strengthen recognition for a certain type of questions, improve the answer given, create new answer and/or recognition around this topic and so on. You will of course need to monitor this change afterwards in order to see if it did help reduce the amount of negative feedback on this answer or not. By regularly monitoring feedback, you can improve the information given to customers.