Explore recipe: Comparing intelligent triage intent predictions with a custom About field

Attention: You must have the Advanced AI add-on to create the report described in this article.

Many customer support operations using Zendesk already encourage their agents or end users to select the theme of the ticket or issue type in a custom ticket field (or fields). These fields are often named something like About, Issue Type, Request Type, and so on. (See The ‘About’ Field.)

This recipe shows you how to compare this type of field against the predicted intent from intelligent triage. The resulting report gives you a better understanding of how the intelligent triage predictions map to the custom field values that you're already used to. In turn, this understanding helps you fine-tune or troubleshoot any automated workflows (like routing, triggers, or views) you create that rely on intelligent triage intent predictions.

This article includes the following topics:

• What you'll need
• Creating the report

What you'll need

Skill level: Intermediate

Time required: 10 minutes

• Zendesk Explore Professional or Enterprise
• Editor or Admin permissions (see Giving users access to Explore)
• Intelligent triage enabled (see Enabling intelligent triage)
• Ticket data in Zendesk Support

Creating the report

1. In Explore, click the reports () icon.
2. In the Reports library, click New report.
3. On the Select a dataset page, click Support > Support - Tickets, then click Start report. The report builder opens.
4. In the Metrics panel, click Add.
5. From the list, select Tickets and click Apply.
6. In the Rows panel, click Add.
7. From the list, select the following attributes and click Apply:
   
   • Intent
   • Intent confidence
   • [Your custom about/issue type field]
8. Click the Intent attribute on the left, go to the Excluded tab, select NULL, and click Apply. This filters your report to show only tickets where the Intent field was populated by the intelligent triage feature.
   Note: If your custom about/issue type field has blank values for some tickets, you can similarly exclude the NULL values by following this same process on your custom field attribute.
9. (Optional) To see more tickets, remove the report’s default date filter or expand the date range.
10. (Optional) Filter the Intent confidence attribute to show only High confidence tickets. Using a High confidence is a good practice for automated processes or deflection.

For ideas on possible use cases once you’ve discovered how your intents are categorized, see Intelligent triage use cases and workflows.

If your workflows involve only routing a ticket to a particular group or agent, then you can instead focus on educating your workforce. Tell them that some tickets might not route properly, and that they can instead correct the routing for the occasion when the prediction didn’t quite match the ticket’s true intent.