Turning Copilot Studio analytics into clear business insight
As organizations scale their use of AI agents, IT teams face a familiar dilemma: How do you give stakeholders the insight they need—without compromising proper oversight?
Expanding access to analytics can mean granting permissions beyond what’s appropriate for a given role. Meanwhile, locking things down preserves control…but creates bottlenecks to information. This forces teams to rely on a small group of people to export and share data across multiple tools just to answer basic questions on analytics—let alone dig into data that truly demonstrates how effective your agents are.
So how can your team get the most relevant, important agent metrics to the right people with the right level of access? Microsoft Copilot Studio now offers two new analytics capabilities that address this need directly:
- Analytics Viewer role, which allows an agent owner to grant read-only access to an agent’s Analytics page.
- Custom metrics, which gives teams the flexibility to add and track specific outcome-based measures aligned to their particular goals for an agent.
The Analytics Viewer role in Copilot Studio
The Analytics Viewer role provides view-only access to an agent’s Analytics page, without permission to edit, configure, publish, or share the agent.
Now generally available, it enables analysts and stakeholders to monitor agent performance securely without affecting agent behavior.
. When you share an agent with Analytics Viewer permissions:
- Users get access only to the Analytics page.
- All other agent pages, actions, settings, test panels, and publishing experiences are hidden and unavailable.
- Analytics Viewers currently can’t define or update custom metrics and savings.
Tip:
For deeper investigation and root-cause analysis, consider granting analysts the Dataverse Bot Transcript Viewer role. This allows Analytics Viewers to drill into the transcripts for detailed analysis.
See the new Analytics Viewer option in the agent sharing experience:
Who should use the Analytics Viewer role in Copilot Studio?
Analytics Viewer is designed for users who need visibility without authoring access. This includes analysts, product owners, business stakeholders, and operations teams that monitor agent performance, trends, and issues.
- This feature supports clear separation of duties: IT pros and admins manage agents, while analysts consume insights and investigate performance.
Analytics page for Analytics Viewers:
Why the Analytics Viewer role matters
For many organizations, access to analytics has been a major blocker for adopting agents in production. It’s understandable. Teams often need production insights, but only a small set of users should be able to change production agents.
Until now, customers have typically exported analytics for offline analysis or built custom reporting outside Copilot Studio. With the Analytics Viewer sharing role, agent owners can share analytics safely—without expanding edit permissions.
Not only does this expand access to insight while maintain proper oversight, it removes a key point of friction, helping teams keep tools streamlined and reduce time to information:
- Stakeholders can access analytics directly, on demand, without relying on exports or intermediaries.
- Teams can grant visibility more broadly, with permissions scoped appropriately to each role.
- Business and product stakeholders can monitor performance that updates after every use, helping them make quicker decisions.
This shift allows analytics to function as a shared, operational capability across teams, while allowing IT and admins to maintain confidence that governance standards are upheld.
“The Analytics Viewer role allows us to provide meaningful performance insights to business and operational stakeholders while maintaining strict production governance. It cleanly separates operational visibility from agent configuration and publishing rights.”Mohamed Arhab
Solution Architect, City of Montreal
Learn more about the Analytics Viewer role here.
Custom metrics: measure the outcomes your business cares about
As agents become part of business workflows, analytics must move from usage to impact. Standard metrics show activity but don’t answer the key question: did the agent achieve the desired business outcome?
Custom metrics in Copilot Studio shift analytics from activity to business outcomes - helping you understand not just usage, but the results agents deliver. They enable evaluation in business terms, supporting ROI, decision making, and stakeholder alignment.
Define success in your own terms
With custom metrics, now available in public preview, you define success in terms your business understands. Custom metrics are calculated by analyzing agent conversations.
You describe:
- The outcome you want to measure.
- The set of allowed result categories that represent that outcome.
These results are classification-based. For example, outcomes might be:
- Ticket deflected or Not deflected.
- Goal completed correctly, Completed with issues, or Not completed.
- Converted or Not converted.
Based on your definition, Copilot Studio automatically generates the prompt that will be used to evaluate agent conversations. This helps keep evaluations consistent, while maintaining the focus on business meaning rather than technical implementation.
Clear, actionable results provided by custom metrics
Each conversation with an agent maps to exactly one outcome category that you define for it. This classification-based approach makes results:
- Easier to interpret, because outcomes are explicit.
- Easier to explain to stakeholders, because they map to business language.
- Easier to act on, because teams can see clear outcome distributions rather than abstract scores.
As examples of clear, actionable results, consider the sample use cases from before:
- Ticket deflection: Understand how effectively (how often, by percentage) an agent resolves issues without escalating to human support.
- Goal completion quality: Evaluate whether users are achieving the intended business outcome successfully (yes/no).
- Sales conversion: Measure how often interactions lead to a qualified outcome or purchase (by percentage or number).
When used thoughtfully, these outcome classifications become a shared language across teams, helping organizations operate agents with greater confidence and clarity as they scale.
How custom metrics help teams operate more effectively
Custom metrics shift agent Analytics pages from being purely a reporting tool to becoming an operational decision-making tool.
Teams can align early on what “good” looks like, using a common definition of success across product and business stakeholders. When performance is evaluated based on outcome distributions—not isolated signals—it’s easier to understand what’s working and what isn’t in your agents.
Over time, this helps facilitate a more intentional approach to agent fine-tuning and iteration:
- Teams can see how changes to an agent affect real outcomes.
- Decisions can be tied directly to business impact.
- Leaders can use measurable results to prioritize investment decisions.
As agents take on more responsibility, this outcome-first model becomes essential for scale because teams need a clear, consistent way to measure impact, align decisions, and confidently expand what agents are trusted to do.
Learn more about custom metrics, now in preview, here.
The right insights for the right people
Individually, the Analytics Viewer role and custom metrics solve different challenges. Together, they help reshape how teams work with analytics and create more value. Analytics Viewer makes it possible to share analytics safely, without expanding access beyond what’s secure and sensible. Custom metrics help ensure your business can measure what you care about.
The result is a better balance between insight and oversight—a fulcrum where visibility can scale across teams without introducing new governance risks. As agents become a core element of business workflows, this combination helps organizations measure impact, demonstrate ROI, and operate agents with greater accuracy and confidence to expand.
That’s the direction Microsoft Copilot Studio continues to move toward: helping organizations not just build and deploy agents, but analyze them with clarity, confidence, and control at scale.
Explore Analytics Viewer and custom metrics in Copilot Studio today to start bringing the right metrics to the right people—at the right level of access.
Microsoft