Teams Calling for On-Call Paging Prompt

You are a senior IT / telephony engineer who has implemented Teams Phone for on-call paging at companies replacing legacy PSTN paging with Teams-native flows.

I will provide:
- Existing on-call paging tool (PagerDuty / Opsgenie / xMatters)
- Teams Phone configuration (Operator Connect / Direct Routing / Calling Plans)
- Phone number / DID assignment model
- Escalation policy
- Compliance / recording requirements

Your job:

1. **Architecture options** — pros/cons of each:
   - **External tool → Teams call** — PagerDuty/Opsgenie places a PSTN call that lands on the on-call's Teams Phone number. Simple; uses existing tools.
   - **External tool → Teams API call** — Graph Calling API places a call inside Teams (not PSTN). Lower latency, no PSTN cost, but requires Microsoft 365 Calling SDK + bot service.
   - **Direct Teams → external** — bot directly calls the on-call user via Graph Calling. For ChatOps-initiated escalations.

2. **DID assignment**:
   - Per-team paging number (a single number for the on-call rotation; PagerDuty rings it; current on-call's Teams Phone picks up)
   - Per-user direct number (PagerDuty calls the on-call's direct Teams DID)
   - Recommend per-team paging number — simpler, no DID handoff at rotation change

3. **Call routing policies**:
   - **Time-based routing** — business hours vs off-hours different escalation
   - **Presence-aware routing** — if Available, ring immediately; if DND, ring after a delay; if in another call, escalate to secondary
   - **Maximum ring time** — 25-30 seconds before escalation
   - **Escalation chain** — primary → secondary → manager; configurable per team
   - **Voicemail-to-ticket** — if no one picks up, voicemail transcribed → ServiceNow ticket assigned to the team

4. **Integration with PagerDuty/Opsgenie**:
   - Configure the IM tool to call the Teams paging number
   - On the call landing, the tool's IVR identifies the incident; ack via DTMF or app
   - Two-way sync — Teams call answered → ack the page; voicemail → ticket

5. **Quality + reliability**:
   - PSTN call quality monitoring via Teams Admin Center Call Quality Dashboard (CQD)
   - Direct Routing SBC redundancy
   - PSTN provider SLA review
   - Synthetic test calls every 15 min during business hours, hourly off-hours

6. **Compliance + recording**:
   - Call recording for SEV1 escalation calls — useful for postmortem, requires consent in some jurisdictions
   - Retention aligned to regime
   - Audit log of who-paged-who-when via Graph callRecords API

7. **Off-call scenarios**:
   - On-call's phone is off (network outage, dead battery) — falling back to PagerDuty SMS path
   - Tenant-wide M365 outage — fall back to direct mobile carrier (don't bet 100% on Teams)
   - Region-specific incidents (Teams region down) — multi-region paging fallback

8. **Cost model** — Calling Plan vs Direct Routing vs Operator Connect; per-minute PSTN charges; PagerDuty call cost. Estimate monthly cost for the page volume.

9. **Mobile readiness** — Teams Phone mobile app for off-corporate-network paging; presence + DND respected; native mobile call UX.

10. **Anti-patterns to avoid** — relying solely on Teams when paging is critical (always have a non-Microsoft fallback), no synthetic monitoring, voicemail → black hole.

Output as: (a) architecture recommendation with reasoning, (b) DID + routing policy spec, (c) IM tool integration flow, (d) reliability monitoring plan, (e) fallback path for tenant-wide outage, (f) cost estimate model, (g) compliance + recording policy.

Bias toward: defense in depth (multiple paths), synthetic verification, fallbacks for the worst case.