Note
Access to this page requires authorization. You can try signing in or changing directories.
Access to this page requires authorization. You can try changing directories.
Copilot Studio has built-in Conversational boosting and Fallback system topics. These topics trigger when the Natural Language Understanding (NLU) model can't find a matching topic or action for a given user query. In terms of priority, Conversational boosting triggers before the Fallback topic.
If most unrecognized utterances go to a human representative, you can improve deflection by addressing usage patterns that consistently trigger Fallback.
Tip
Topic enrichment is an offline data analysis exercise, focused on repurposing user queries that triggered the Fallback topic into triggering relevant topics in Copilot Studio.
The analyzed user queries from the Fallback topic typically fall into these buckets:
User queries that are expected to trigger existing topics, but the agent's NLU missed them.
User queries that can be converted to newly suggested topics.
Unmapped user queries that aren't relevant to any existing or new topics.
Other categories, including:
- User queries that triggered a Multiple Topics Matched (also known as "did you mean") topic followed by Conversational boosting or Fallback.
- Unclear user queries that hit Conversational boosting or Fallback.
- User queries from incomplete conversations that led to Conversational boosting or Fallback.
Of these categories, the first two 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.
