Topic confusion analysis
Using semantically similar trigger phrases for two different topics can lead to confusion; the copilot might not know which topic to open, and needs to ask follow-up questions to the end-user.
Tip
Confusion Is typically measured by the frequency of the Multiple Topics Matched topic (also known as "did you mean") being triggered.
This often results in the escalation to an agent, and spikes the deflection rate of the copilot.
A topic confusion analysis exercise helps you improve topic triggering accuracy by finding overlaps between topics. Resolving topic overlaps can help reduce the need for the copilot to ask clarifying questions before triggering a topic.
Identifying semantically similar trigger phrases can also help you determine if you have topics that themselves are similar and could be consolidated to simplify the copilot authoring process, or edited to make the topics more distinct with high triggering accuracy, which in turn improves the deflection rate.
You can identify the list of topics causing confusion during triggering, by enabling the AI features for Teams and Classic copilots in Copilot Studio for topic overlap detection.
This standard capability helps you identify the trigger phrases causing confusion, remove duplicates, and consolidate similar topics.
Important
Only classic chatbots or Teams copilots have access to these legacy AI capabilities.
