Communicate the value story to stakeholders

Metrics become a business case when they reach the right audience in a form they can understand. This article covers three measurement tiers in Copilot Studio, the Microsoft surfaces that extend those tiers, the value curve over time, a five-step operating rhythm, and a one-page scorecard template for your next executive review.

Measurement tiers

Start with Copilot Studio analytics

Every agent you publish in Copilot Studio emits analytics from the first conversation. Copilot Studio Analytics documents all built-in key performance indicators (KPIs).

Conversational agent KPIs

Conversational agent KPIs tell you whether the agent is being used to your expectations, including whether it's resolving user questions or requests.

Learn more: Analyze conversational agents

Autonomous agent KPIs

For agents that run on event triggers rather than user chat, Copilot Studio analytics reports run outcomes, trigger use, tool use, and knowledge source use. Together, these metrics tell you whether the autonomous agent is doing what you designed it to do.

Learn more: Analyze autonomous agents

Quality and voice-of-customer signals

• Reactions: Thumbs-up and thumbs-down reactions, plus free-text comments, retained for 28 days.
• CSAT: End of Conversation survey, scored 1 to 5.
• Sentiment: AI-scored negative-sentiment percentage across sessions.
• Themes: AI-grouped user questions that expose topic-coverage gaps, with one-select creation of evaluation test sets from any theme.
• Generated answer rate and quality: Surfaces unanswered questions and response quality for remediation.
• AI-generated insights: Daily recommendations on unanswered generative questions with specific coverage fixes.

Learn more: Analyze conversational agents

Savings calculator

Use the Savings calculator on your agent's Analytics page to enter the estimated time or money saved per run or per tool. Copilot Studio computes totals in real time against successful runs in the period, and it updates retroactively when you change inputs. You can produce a defensible return on investment (ROI) tile with no data engineering.

Custom metrics

Define up to three custom metrics per agent by using natural language. Copilot Studio generates a prompt that scores a sampling of sessions and displays the metric as a labeled donut graph under Custom metrics on the Analytics page.

Set up Copilot Studio Kit for quality and governance

The Copilot Studio Kit is a solution that you can install to extend Copilot Studio functionality with testing, governance, and cross-agent reporting. Review the Conversation Analyzer and Conversation KPIs features to extend your analytics strategy.

Use Azure Application Insights for deep telemetry

For agents that participate in a broader business process, Azure Application Insights captures structured traces, dependencies, custom business events, and end-to-end latency. Connect from the agent's Settings > Analytics pane. Once connected, the agent emits logged messages, topic-trigger events, and any custom telemetry events your topics produce.

Sample KQL: sessions and messages per day for the last 30 days

This sample Kusto Query Language (KQL) query counts sessions and messages per day for the last 30 days, which is a common starting point for understanding adoption and usage patterns.

requests
| where timestamp > ago(30d)
| summarize sessions = dcount(session_Id), messages = count() by bin(timestamp, 1d)
| render timechart

The same pattern supports questions about top topics, escalation rate over time, agent-level latency, error recurrence, and peak-hour volume. Learn more in Capture telemetry with Application Insights.

Extend telemetry with Microsoft reporting tools

After you establish baseline analytics in Copilot Studio, Copilot Studio Kit, and Azure Application Insights, use Microsoft reporting tools like Viva Insights and Copilot Control System reporting to connect usage, quality, sentiment, and business outcomes.

Review your Copilot Studio agents report in Viva Insights

The Copilot Studio agents report is a Power BI template you run from the Viva Insights analyst experience. The report combines all agents in your tenant into a single set of dashboards: Agent Overview, Top agents used, Conversational Agent Impact, Autonomous Agent Impact, and the customizable Agent Assisted Hours and Agent Assisted Value cards.

In addition to the metrics in Copilot Studio Analytics, the Copilot Studio agents report surfaces:

• Agent Assisted Hours and Agent Assisted Value computed against the Microsoft-published formula, with a calculator you can override to reflect your own productive-hour rate
• Cross-agent rollups so you can compare 10 agents at once, rather than reading 10 separate Analytics pages
• Trend charts on engagement, satisfaction, and resolution that go further back than the native 30-day window in Copilot Studio Analytics
• Side-by-side conversational and autonomous metrics, which is helpful when your portfolio mixes both

Connect Copilot usage to business outcomes with the Copilot business impact report

Check how Copilot usage affects the business outcome metrics your organization already tracks by using the Copilot business impact report. These metrics include deals closed, cases resolved, on-time delivery, and claims processed. Upload your own outcome data to Viva Insights. The report joins that data with Copilot usage, so you can understand the relationship and identify where increased adoption would realize the most outcome value.

To use this report effectively:

• Identify the Copilot usage behaviors most associated with favorable business outcomes, such as intensity, breadth of features, and consistency.
• Find teams or functions with the lowest adoption and the largest outcome gaps so you can target your enablement budget.
• Quantify untapped value or how much more Agent Assisted Value is available if low usage cohorts move to high usage.
• Bring the resulting analysis into your executive scorecard so the value story stays grounded in your own organizational metrics, not generic benchmarks.

Track adoption and impact in the Copilot Dashboard

The Copilot Dashboard in Viva Insights is the strategic adoption and impact surface for Microsoft 365 Copilot and the agents that extend it. The dashboard has three tabs your sponsor cares about: Adoption (who is using Copilot, how often, on which surfaces), Impact (workplace patterns, sentiment from Pulse and Glint, and Copilot Assisted Hours), and Value (the dollar value of those assisted hours).

Use the Copilot Dashboard to give your sponsor a regular readout on how Copilot is changing work in your organization. The dashboard provides:

• Routine adoption rate and usage intensity at the function level, so you can see which teams are pulling ahead and which need help.
• Workplace pattern shifts, such as meeting hours, focus time, and after-hours work, that show how Copilot is changing the shape of work, not just the volume.
• Productivity-impact estimates that translate Copilot actions into hours and dollars, with a built-in calculator you can override to reflect your own productive-hour rate.
• Industry-benchmark comparisons that give you context for whether your adoption is on pace, ahead of, or behind similar organizations.

Capture how people feel with the Viva Glint Copilot Impact Survey

The Viva Glint Copilot Impact Survey is the Microsoft-published, Glint-native survey template for measuring how Copilot is changing your employees' experience of work. The four core items (productivity, speed, effort, quality) are standardized and feed sentiment signals back into the Copilot Dashboard, so you can correlate sentiment with adoption and impact in one place.

To use this survey effectively:

• Establish a sentiment baseline before broad rollout, and run the survey on a recurring cadence to see the trend, not a single point.
• Pair the survey with your usage data so you can identify cohorts where high adoption coexists with low sentiment, which often points to a training or workflow gap.
• Surface the open-ended comments in your executive scorecard's Voice of the User section, alongside CSAT and Conversation Analyzer themes.

Tie it all together with Copilot Control System measurement and reporting

The Copilot Control System is a measurement and reporting framework. It brings operational reports in the Microsoft 365 admin center, strategic reports in the Copilot Dashboard, and customizable reports in Viva Insights into one view of readiness, adoption, productivity impact, and business value. When stakeholders ask for one place to see the full story, point them to this resource.

The Copilot Control System provides:

• Operational reports for license assignments, agent billing, and deployment health (Microsoft 365 admin center).
• Strategic adoption and impact reports for sponsors and HR (Copilot Dashboard).
• Customizable analytics for analysts who want to combine Copilot data with their own organizational metrics (Viva Insights, the Copilot Studio agents report, and the Copilot business impact report).
• A single governance lens, so security, management, and measurement pillars are seen as one system rather than three separate efforts.

How value grows in stages

Use this table to understand what to expect at each stage of adoption, from early efficiency gains to long-term strategic value.

Five steps to build measurement discipline

Use these five steps to build measurement into the full agent lifecycle, from planning through ongoing review.

1. Before you build: Define what value looks like and align on one or two outcomes the agent moves.
2. Before go-live: Document today's state. Three numbers—time per task, error rate, cost per interaction—captured before launch carry the entire measurement story.
3. At launch: Turn on measurement from day one by enabling Copilot Studio Analytics, connecting Application Insights, and installing the Copilot Studio Kit. Every day of missing telemetry is a gap you can't recover.
4. At 30 and 90 days: Connect the data to your value pillars, compare movement against baseline, and name the pillar explicitly in every review.
5. Ongoing: Make ROI a rhythm rather than a report. Schedule the 90-day review before the agent goes live. The teams that measure consistently are the teams that discover expansion opportunities.

Build your executive scorecard

Metrics on a dashboard aren't the same as metrics on a scorecard. A scorecard tells one story on one page. Use this structure for your quarterly value review.

Conclusion

Organizations that capture real business value from AI agents treat measurement as a first-class product artifact. They define success before they build, capture a telemetry baseline, instrument every agent on day one, review results on a regular cadence with a named sponsor, and communicate outcomes in the language executives use.

Your measurement stack starts with Copilot Studio and Copilot Studio analytics, extends through Copilot Studio Kit and Azure Application Insights, and ties together with the Copilot business impact report, the Copilot Dashboard, the Viva Glint Copilot Impact Survey, and the Copilot Control System measurement and reporting framework.

When you get this approach right, agents become infrastructure, and people reinvest time in judgment, creativity, and customer relationships—which is exactly where they add the most value.

Next steps

Two reference articles support the work in this series.

• Agent metrics reference defines every metric named across all four articles, explains each metric in plain English, shows the use cases it applies to, and identifies the Microsoft surface where it appears.
• Use case blueprints for measuring agent value provides worked examples across 16 common business functions. Each example includes realistic inputs and step-by-step Agent Assisted Hours calculations that you can adapt for your own agent.