Question

A team has built a customer-support agent in Microsoft Copilot Studio. Before expanding it to more channels, managers want a systematic way to measure how accurately the agent's responses match expected answers across a set of representative test questions. Which Copilot Studio capability should they use?

Accepted Answer

Agent evaluations, which run test questions against expected results to assess response quality

Answer

Analytics dashboards that show the total number of sessions and engagement rates

Answer

Publishing the agent to additional channels to gather more live feedback

Answer

Connecting more knowledge sources to broaden the agent's coverage