Analytics-Ready Playbook

Custom dashboards and BI integration in half a day

---

Prerequisites

• Quick Start playbook completed — baseline KPIs and review cadence in place
• Personal Access Token (PAT) with manage_billing:copilot scope
• A BI or visualization tool — Power BI, Tableau, Jupyter, Grafana, or similar

PAT Security

Store your PAT securely. Never commit it to source control. Use a secret manager or environment variable.

---

Step 1: Set Up API Access

Create a PAT

1. Settings → Developer settings → Personal access tokens → Fine-grained tokens.
2. Set resource owner to your enterprise.
3. Enable manage_billing:copilot (read).
4. Generate and copy the token.

Test Your Access

curl -s -L \
  -H "Accept: application/vnd.github+json" \
  -H "Authorization: Bearer YOUR_PAT" \
  -H "X-GitHub-Api-Version: 2022-11-28" \
  "https://api.github.com/enterprises/ENTERPRISE/copilot/usage" \
  | head -c 500

Expected Result

NDJSON output with day, total_active_users, total_engaged_users, and breakdown arrays. A 401/403 means incorrect PAT scopes or enterprise slug.

---

Step 2: Automate Data Collection

Replace manual exports with a scheduled daily job.

GitHub ActionsScheduled Script (cron / Task Scheduler)

Create .github/workflows/copilot-metrics-collect.yml:

name: Collect Copilot Metrics
on:
  schedule:
    - cron: '0 6 * * *'  # Daily at 6 AM UTC
  workflow_dispatch: {}

jobs:
  collect:
    runs-on: ubuntu-latest
    steps:
      - name: Pull Copilot usage data
        env:
          GH_TOKEN: ${{ secrets.COPILOT_METRICS_PAT }}
          ENTERPRISE: your-enterprise-slug
        run: |
          DATE=$(date -u +%Y-%m-%d)
          curl -s -L \
            -H "Accept: application/vnd.github+json" \
            -H "Authorization: Bearer $GH_TOKEN" \
            -H "X-GitHub-Api-Version: 2022-11-28" \
            "https://api.github.com/enterprises/$ENTERPRISE/copilot/usage" \
            -o "copilot-usage-${DATE}.ndjson"

      - name: Upload artifact
        uses: actions/upload-artifact@v4
        with:
          name: copilot-metrics-${{ github.run_id }}
          path: copilot-usage-*.ndjson
          retention-days: 90

#!/bin/bash
DATE=$(date -u +%Y-%m-%d)
OUTDIR="/data/copilot-metrics"
mkdir -p "$OUTDIR"

curl -s -L \
  -H "Accept: application/vnd.github+json" \
  -H "Authorization: Bearer $COPILOT_METRICS_PAT" \
  -H "X-GitHub-Api-Version: 2022-11-28" \
  "https://api.github.com/enterprises/$ENTERPRISE/copilot/usage" \
  -o "$OUTDIR/copilot-usage-${DATE}.ndjson"

echo "[$(date)] Collected metrics for $DATE" >> "$OUTDIR/collection.log"

Schedule with cron: 0 6 * * * /path/to/collect-copilot-metrics.sh

Data Retention

The API returns up to 28 days of history. Set up alerting on your job to catch failures before gaps form.

---

Step 3: Choose a Visualization Tool

Tool	Best For	Effort	Customization
Power BI / Tableau	Enterprise BI teams	Medium	Full
copilot-metrics-viewer	Quick standalone dashboard	Low	Moderate
Jupyter / Python	Data science, ad-hoc exploration	Low	Full
Grafana	Teams already on Grafana	Medium	Full

All tools ingest NDJSON — pick what your team already knows.

---

Step 4: Build Your First Dashboard

Include these four panels in your initial dashboard.

Panel 1: DAU / WAU / MAU Trend

Line chart — Daily, 7-day rolling, and 28-day rolling active users. Primary adoption health signal.

Panel 2: Acceptance Rate by Language

Stacked bar — Acceptance rate per language. Low rates flag training or prompt-quality opportunities.

Panel 3: Seat Utilization

Gauge / KPI card — Active ÷ assigned seats (target: 70%). Show inactive seat count (no activity in 30 days).

Panel 4: Feature Mix Over Time

Stacked area — Completions, chat, CLI, PR summaries. Broadening mix = deepening adoption.

Sample Dashboard Layout

┌─────────────────────────────────┬──────────────────────┐
│                                 │                      │
│   DAU / WAU / MAU Trend         │  Seat Utilization    │
│   (line chart, full width)      │  (gauge + KPI card)  │
│                                 │                      │
├─────────────────────────────────┼──────────────────────┤
│                                 │                      │
│   Acceptance Rate by Language   │  Feature Mix Over    │
│   (stacked bar chart)           │  Time (stacked area) │
│                                 │                      │
└─────────────────────────────────┴──────────────────────┘

---

Step 5: Map Stakeholders to Metrics

Stakeholder	Key Metrics	Cadence
Engineering Managers	DAU/MAU, acceptance rate, feature mix	Weekly
VP of Engineering	Seat utilization, adoption trend, cost per active user	Monthly
CFO / Finance	Seat utilization, cost per active user, inactive seats	Quarterly
CISO / Security	Policy enablement, allowed/blocked features	Quarterly
Developers	Personal acceptance rate, language-specific quality	On demand

Tailor the Message

Create filtered views per audience — managers want team detail; finance wants cost efficiency.

---

Step 6: Establish Reporting Cadence

Weekly Ops Review (15 min)

Audience: Engineering managers, platform team

• DAU trend, acceptance rate changes, new feature adoption
• Action: Address adoption blockers

Monthly Executive Summary (30 min)

Audience: VP of Engineering, department heads

• MoM adoption growth, seat utilization, top 3 wins
• Action: Adjust seat allocation, plan enablement

Quarterly Business Review (60 min)

Audience: CFO, CTO, CISO

• Quarterly trajectory, cost per active user, ROI estimate (hours saved × blended rate), compliance posture
• Action: Budget decisions, policy updates, expansion plans

---

Result: Automated daily collection, a four-panel dashboard, stakeholder-aligned reporting from weekly ops to quarterly QBR.

---

When to Level Up

You've outgrown this playbook when leadership asks "Is Copilot making us ship faster?" and you need to correlate usage with DORA metrics or build an ROI narrative for budget renewal.

Move to Using Apache DevLake if you want a prebuilt correlation stack.

---

What to Do Next

• Set up automated data collection today
• Build your first dashboard with the four panels above
• Schedule a stakeholder alignment meeting on metrics and cadence
• Read Using Apache DevLake when you want a prebuilt path to prove impact