Hours saved · assisted value · adoption & readiness — a defensible ROI narrative aligned to Microsoft's Frontier Firm framework.
Found this useful? ⭐ Star this repo to help others discover it!
Deployment paths ↓ · What it measures ↓ · Data sources ↓ · Dashboard pages ↓ · Research ↓
Both play here in the page — no download.
Demo — what the dashboard measures, page by page (1m 49s)
ValueLens-Demo.mp4
Setup guide — getting your own data in, every source, start to finish (7m 07s)
ValueLens-Setup.mp4
👉 New here? Start with 1. Local CSV
It ships with a fabricated sample dataset that fills the whole dashboard. Open the template, point it at three CSVs, done — no tenant, no exports, no setup. Roughly two minutes, and it tells you whether the numbers are worth wiring up before you wire anything up.
⚠️ Usage & compliance disclaimer
While this tool helps customers understand the business value of their AI usage data, Microsoft has no visibility into the data customers input, nor control over how the template is used. Customers are solely responsible for ensuring their use complies with all applicable laws and regulations (including data privacy and security). Microsoft disclaims all liability arising from use of this template.
This is an experimental template with Purview audit logs as the primary source. Audit logs provide visibility into Copilot/agent interactions but are not intended as the sole source of truth for licensing or full‑fidelity reporting. Not supported through Microsoft support channels — please open an issue in this repo.
Three paths, same dashboard. They produce identical pages and the same value model — they differ only in how the data gets in and refreshes. Work down the table: each one is a step up in effort and a step up in automation.
| Path | Pick this when… | Setup |
|---|---|---|
| 1. Local CSV · start here 🧪 | You want to see it working now — or run a one-off look at your own numbers. | Sample data included. Open the template, point it at three CSVs. No tenant, no exports, no scripts. Then swap in your own export when ready. |
| 2. SharePoint · scheduled, Pro | You want it refreshing on its own on Power BI Pro — no Fabric or Premium. | A scheduled script extracts, rolls up and uploads to SharePoint; Power BI refreshes on a timer. |
| 3. Fabric · scale · recommended | You have Fabric capacity (or Premium / PPU), or any Spark + SQL stack. | Notebooks shape the data into a Lakehouse — best performance, plus the optional billing & feedback pages. Also runs on Databricks, Synapse or Azure SQL. |
Not sure? Start with path 1. It takes minutes and tells you whether the numbers are worth automating — before you set up any automation. Move to 2 or 3 when you want it hands-off.
Running Copilot Studio agents and want the deeper agent / topic / CSAT pages? That's an optional add-on inside path 3 — see
3. Fabric/extended/. Stand up path 3 first. Landing agent transcripts in Dataverse instead? Use the Dataverse companion repo ↗, which reads them natively — no Fabric or SharePoint needed.
Each path folder has its own README with the exact, step‑by‑step setup. This page is just the map.
Both are at the top of this page and play inline.
- Demo (1m 49s) — a tour of the pages and how the value model fits together.
- Setup guide (7m 07s) — every data source (Purview audit, Entra, M365 usage), the permissions each needs, the parameters you fill in, and the Fabric path in full: PySpark notebooks instead of scripts, the Graph permissions per notebook, the ingester-then-processor run order, and the same notebooks and template running on Databricks, Synapse or Azure SQL.
📁 Repo layout
README.md · LICENSE · Images/
1. Local CSV/ Local CSV.pbit · sample-data/ ← start here, fabricated demo dataset
2. SharePoint/ SharePoint.pbit · scripts/ · azure-container/
3. Fabric/ Fabric.pbit · docs/ · flows/ · notebooks/ · pipelines/
extended/ optional Copilot Studio add-on — deeper agent, topic/CSAT & PPAC credit pages
archive/ superseded versions — kept for reference, not maintained
Dataverse path → companion repo: Keithland89/Copilot-Studio-Agent-Analytics
- Quantified value — hours saved and dollar‑equivalent assisted value, grounded in research‑sourced time baselines.
- Value by function — Sales, HR, IT, Legal, Finance, Marketing, Customer Service, with task‑level attribution.
- User maturity — Beginner → Developing → Power, from usage breadth and agent adoption.
- Business case — projected annual value, ROI multiple, and licence investment net.
How: every interaction → classified into an AI Task → mapped to a research‑sourced time baseline → summed to Hours Saved → × hourly rate = Assisted Value.
| Source | Required? | Where it comes from |
|---|---|---|
| Copilot interactions (audit logs) | ✅ Core | Microsoft Purview |
| Licensed users | ✅ Core | Microsoft 365 Admin Center |
| Org data (department / function) | ✅ Core | Microsoft Entra |
| Agents 365 | ⬜ Optional | Agent 365 export (Fabric path) |
| Cowork / Work IQ consumption | ⬜ Optional | Microsoft 365 Admin Center export → see 3. Fabric/flows/COST-CONSUMPTION.md |
| Credit consumption (billing) | ⬜ Optional | Power Platform Admin Center export → see 3. Fabric/extended/Fabric + Copilot Studio/CREDIT-CONSUMPTION-SETUP.md (Studio add-on) |
| Product feedback | ⬜ Optional | M365 Admin Center → Health → Product Feedback export |
| Copilot Studio agent transcripts | ⬜ Optional | Dataverse ConversationTranscript table — use the Dataverse companion repo ↗ |
Optional sources are gated by Enable_* toggles — the dashboard works fine without them. The exact
export + connect steps live in the path README you choose above.
14 core pages (13 on SharePoint) + 3 Copilot Studio add-on pages
Page names below match the actual report tabs. The ✓ columns show which deployment path ships each page — see the path README for that build's own list.
| Page | SharePoint | Fabric | Purpose |
|---|---|---|---|
| ◆ Activation | ✓ | ✓ | Activation across teams — licensed vs unlicensed, active vs inactive |
| 🎯 Readiness | ✓ | ✓ | Ranks unlicensed / low-adoption users by upgrade‑priority score |
| 📡 Adoption | ✓ | ✓ | User counts, coverage %, licensed vs unlicensed reach |
| 🪙 Consumption ¹ | ✓ | ✓ | Copilot & agent consumption — credits / messages over time |
| 🔮 Activity | ✓ | ✓ | Copilot and agent usage, tasks and behaviour mix |
| 🚀 Value | ✓ | ✓ | Hours saved, dollar‑equivalent assisted value, and the business case |
| 🌱 Maturity | ✓ | ✓ | Progression: Asking → Finding → Consuming → Producing → Delegating |
| 🛡 Agent Health | ✓ | ✓ | Agent resolution, abandonment, escalation and response time |
| 💬 Feedback | — | ✓ | Thumbs up/down sentiment and verbatim feedback themes |
| 📈 Heatmap | ✓ | ✓ | Activity heatmap across the reporting period |
| 🏅 Leaderboard | ✓ | ✓ | Top users, agents, and functions |
| 📘 Appendix: Key Concepts | ✓ | ✓ | Methodology and key‑concept explainers |
| 🧬 Appendix: Signal Table | ✓ | ✓ | Trace raw signals through to value (audit trail) |
| 📘 Appendix: Glossary | ✓ | ✓ | Metric definitions and research sources |
¹ Labelled 🪙 Consumption in the SharePoint template and 🪙 Credit Meter in the Fabric template — same page, different tab name.
Copilot Studio add-on — three extra pages in the Fabric + Copilot Studio build:
| Page | Purpose |
|---|---|
| Copilot Studio: Credits Consumed | Agent credit consumption and billing breakdown |
| Copilot Studio: Agent Evaluation | Agent resolution, abandonment, escalation and response time |
| Copilot Studio: Topic Analysis | Most‑asked topics, resolution and abandonment by agent |
Published time-baseline sources behind the value model
Human‑time baselines are drawn from published research — Microsoft Research, MIT/Science (Noy & Zhang 2023), NBER (Brynjolfsson et al. 2023), BCG/Harvard (Dell'Acqua et al. 2023), McKinsey, Forrester TEI, IDC, and others. The full per‑task source list is in the 📖 Metric Glossary page inside the template.
Built by the Microsoft Copilot Growth & ROI practice, building on the structure of the community AI‑in‑One Dashboard. Licensed MIT — see LICENSE.