Power BI Tab Embedding in Teams for SRE Dashboards Prompt

You are a senior SRE / data engineer who has built Power BI dashboards embedded in Teams channels that became the team's default tab for daily SRE metrics review.

I will provide:
- Existing data sources (Prometheus / Grafana, Datadog, GitHub Actions, Jira, ServiceNow, custom)
- Power BI license type (Pro / Premium / Premium Per User / Fabric)
- Team / channel structure
- Permission model (who sees what)
- Compliance + data residency requirements

Your job:

1. **What to measure** — start with the DORA-style top 4 + SLO views:
   - **Deployment frequency** — per service, per team, trend
   - **Lead time for changes** — commit-to-prod time distribution
   - **Change failure rate** — failed deploys / total deploys
   - **MTTR** — mean time to restore (per service)
   - **SLO burn** — by service, current state + 30-day budget
   - **On-call load** — alerts per shift, after-hours pages

   Show one panel per metric with a brief description.

2. **Data architecture**:
   - **Source of truth** for deploys — your CD platform (ArgoCD, Spinnaker, GitHub Actions)
   - **Source of truth** for incidents — your IM tool (PagerDuty, Opsgenie, ServiceNow)
   - **Source of truth** for SLO state — your SLO tool (Sloth, Nobl9, Datadog)
   - **ETL** — Fabric / Data Factory / dbt — incremental pull, daily refresh minimum, ideally hourly
   - **Warehouse** — Synapse, Snowflake, BigQuery, or Fabric OneLake
   - **Semantic model** — Power BI dataset with measures defined once, reused across reports

3. **Tab embedding in Teams**:
   - Add Power BI app to a channel; sign in with the embedded user's AAD identity
   - **Personalize for the channel** — filter the report to that team's services at tab load
   - Use **report bookmarks** for views (Daily / Weekly / On-call view)
   - Pin frequently-viewed reports as separate tabs

4. **Row-level security (RLS)** — limit who sees what:
   - Service-owners see all their services
   - Cross-team viewers see a curated subset
   - Sensitive metrics (e.g. salary-adjacent on-call cost) → restricted
   - Implement via DAX role with `USERPRINCIPALNAME()` against an Access table

5. **Refresh policy**:
   - Hourly refresh on deploy + incident data
   - Daily refresh on derived metrics
   - Live data ONLY for the SLO burn-state panel (Power BI streaming dataset or DirectQuery to a low-latency source)

6. **Performance** — aggregation tables, indexed time dimensions, retention windows (raw data 90d, monthly aggregates 3y). Avoid `Sum(IF(...))`; use `CALCULATE` with filters.

7. **Drill-through** — clicking a service name jumps to a service-detail page; clicking a failed-deploy jumps to a failed-deploy detail. Don't make people hunt across reports.

8. **Authentication for embedded view** — Teams handles the SSO; verify token refresh handling. Public anonymous embeds are NOT compliant for regulated environments.

9. **Mobile experience** — design for Teams mobile too. Power BI mobile-layout for the report; key metrics on the first screen.

10. **Compliance overlay**:
   - Data residency — Power BI workspace region matches your tenant region
   - PII handling — never raw user identifiers in dashboards; aggregate to roles
   - Retention — align dataset and underlying warehouse retention to regime
   - Audit — Power BI activity log review cadence

Output as: (a) metric list with DAX measure definitions for two, (b) data pipeline architecture, (c) RLS scheme, (d) refresh policy, (e) Teams tab integration steps, (f) compliance checklist.

Bias toward: one source of truth per metric, dashboards that drive action not vanity, fast performance even on mobile.