Share via

Topic enrichment analysis

  • Article

Important

Power Virtual Agents capabilities and features are now part of Microsoft Copilot Studio following significant investments in generative AI and enhanced integrations across Microsoft Copilot.

Some articles and screenshots may refer to Power Virtual Agents while we update documentation and training content.

Analyzing the Fallback topic conversations

Microsoft Copilot Studio has a built-in Fallback system topic set to trigger when the Natural Language Understanding (NLU) model isn't able to find a matching topic for a given user query.

If most Fallback topic conversations are escalated to a human agent–that is, not deflected–there's an opportunity to improve the deflection by addressing the usage patterns of the user that triggers fallback consistently.

Tip

Topic enrichment is an offline data analysis exercise, focused on repurposing user queries that triggered the Fallback topic into triggering relevant topics in Microsoft Copilot Studio.

The analyzed user queries from the Fallback topic typically fall into these buckets:

  1. User queries that are expected to trigger existing topics, but somehow are missed by the copilot's NLU.

  2. User queries that can be converted to newly suggested topics.

  3. Unmapped user queries that aren't relevant to any existing or new topics.

  4. Other categories, including user queries that triggered a Multiple Topics Matched (also known as "did you mean") topic followed by Fallback, unclear user queries that hit Fallback, and user queries from incomplete conversations that led to Fallback.

Of the four categories, the first and the second are immediately actionable. Based on the findings from these categories, you can enrich the topics by adding more trigger phrases for existing topics or creating new topics.

Diagram that illustrates a process flow to improve fallback analysis.

Topic enrichment through out-of-the-box analytics

Microsoft Copilot Studio provides advanced AI capabilities out of the box. To identify a list of suggested new topics, turn on the advanced AI features for topic suggestion from chat transcripts, when the author doesn't want to enable the Fallback topic.

This information can also be used to create new topics to improve deflection rate.

Screenshot of the topic suggestions from chat transcripts window.

Topic confusion analysis