Adatum Health: Three Countries, Three PSTN Answers, One Extension Plan
Adatum Health has 2,500 Teams Phone users: 1,200 in the US, 800 in the UK, 500 in Germany. The US has a pair of certified SBCs in a Dallas datacenter with three years left on a SIP contract and 1,200 DIDs. The UK has no on-premises telephony and must port 800 numbers from a local carrier within 90 days. Germany requires call media and signaling to remain in the EU, and its carrier will not peer with a cloud SIP trunk. A 220-agent contact center needs a multi-level auto attendant, 12 call queues, holiday routing, and supervisor silent monitoring. One 150-user clinic site has a 40 Mbps MPLS link with 300 ms round-trip latency to Dallas and cannot get local internet. All users are M365 E3; agents have no add-on. Leadership wants a 'one number strategy' with four-digit extension dialing across all three countries.
1. Select the PSTN connectivity option for each country and explain what would break if Adatum standardized on one option.
Read the worked answer
Match the option to the constraint that cannot be moved.
- US - Direct Routing. The certified SBC pair, the three-year SIP contract, and 1,200 DIDs already exist. Replacing them discards sunk cost and forces number porting for 1,200 DIDs with no benefit. Keep the SBCs, configure survivability (paired SBCs, failover to the second, documented Teams SIP proxy FQDNs) and add voice routing policies.
- UK - Operator Connect (with Calling Plans as fallback). No SBC, no telephony staff, and a hard 90-day port deadline. A carrier-managed Operator Connect arrangement transfers porting risk to an accredited operator. Calling Plans are the fallback if no operator can port those specific numbers in time. Direct Routing is the worst choice here - building and certifying an SBC in 90 days is the delay, not the port.
- Germany - Direct Routing with SBCs in a Frankfurt datacenter. Two constraints stack: in-country media/signaling and a carrier that will not peer with a cloud trunk. Operator Connect cannot be used against a carrier that will not peer, and Calling Plans would place PSTN outside the residency boundary.
Standardizing breaks: one option everywhere ignores the US contract, the UK deadline, and the German carrier and residency constraints. Standardization is a management aspiration; the constraint model is the design. Your deliverable is a per-country solution with one consistent dial plan and routing model on top.
2. Decide what to do about the clinic site (40 Mbps, 300 ms RTT to Dallas, no local internet).
Read the worked answer
Two separate problems, two separate answers.
- Health of the phone system when the WAN fails. This is what a survivable branch appliance addresses: a certified SBC with SBA at the site registers fallback users and keeps PSTN dial tone local if the link to Dallas drops. It is insurance, not performance.
- Media quality for Teams calls and meetings. 300 ms round-trip latency is far outside what real-time media tolerates - it will produce talk-over and degraded audio on every call regardless of how the PSTN leg is delivered, and 40 Mbps shared by 150 users limits concurrent video. An SBA does not fix this.
Recommendation:
- Pursue a local internet circuit (business broadband or 5G/LTE failover) with the egress funded as a clinic-critical service. Without it, the clinic will generate a steady complaint stream that looks like a Teams problem and is a network problem.
- In parallel, deploy the local SBC/SBA for PSTN survivability and keep analog lines for critical clinical handsets.
- Until the circuit exists, enforce a restrictive meeting policy for the clinic (video off by default for external calls, audio-only for groups, certified headsets), and document it as a time-boxed exception with a review date.
Do not silently cut the clinic over and wait for complaints - a clinical site's degraded phone quality is a safety issue, not a satisfaction metric.
3. Design the contact center and the four-digit dialing model. What can be done natively and what needs to be purchased?
Read the worked answer
Native capabilities: auto attendants with multi-level call flow, call queues with routing and overflow, after-hours call flows, and queue/agent licensing via resource accounts. Each auto attendant and queue needs its own resource account and a phone number license; forgetting this is the classic cutover blocker. Holiday routing needs verification against the current product limits - the practical fallback is per-date custom handling rather than assuming a full holiday calendar exists.
Not native, needs budget: supervisor silent monitoring and barge are not built-in Teams Phone features; they require a certified contact center partner application or advanced call-queue capabilities tied to an add-on. Skills-based routing and similar advanced queue behaviors also typically require Teams Premium for the agents or organizer - verify before promising it to the business. Agents currently have no add-on, so this is a procurement task in week one, not a configuration task in week ten.
Four-digit dialing across countries:
- Assign non-overlapping extension blocks per country (e.g., 1xxx US, 2xxx UK, 3xxx DE). Overlapping ranges across countries is the failure mode nobody catches until two people dial the same extension.
- Use tenant- and site-level dial plans with normalization rules that convert a four-digit match to the user's E.164 number, so internal calls stay inside Teams with no PSTN cost.
- Put the ranges in the employee directory; a dial plan that exists but isn't published still generates 'wrong number' tickets.
- Validate by calling across all three countries plus one external PSTN call from each, and verify the contact center queue does not strip the extension before routing.