What happens when someone dials 911 from a desk phone at a second site?

By Michael Obraitis, Managing Partner, and Michael Dale, CEO · Published August 17, 2026 · Last reviewed August 2026 · ADAM Pulse Knowledge Base · Carrier Transition Series

Short answer

The call is resolved by where the phone is, not by who the user is — provided the configuration actually knows where the phone is.

That proviso is the entire article. A branch desk phone works perfectly for ordinary calls whether or not its emergency location is right, so nothing tells you the location is wrong until it matters.

The resolution order applied to a branch desk phone, with the failure point at each step:

1. Switch MAC and port via LLDP — resolves to building and floor when the port is mapped. Fails if LLDP is off, the switch is not mapped, or the phone was recabled.
2. Wireless BSSID — resolves to the site and floor. Fails on a wired phone, a replaced access point, or an unrecorded BSSID.
3. Private IP or subnet — resolves only if the subnet is unique. Fails when subnets are reused across sites, which is very common.
4. Public IP — resolves only if it is unique to the site. Fails under centralized egress, where every site looks like headquarters.
5. Device address on the phone — whatever was typed in. Fails if never set, or set at headquarters before the phone shipped.
6. Site default emergency address — the site's own default. Fails if the site inherited headquarters' address or was never configured.

Steps 3 through 6 are the dangerous ones. They resolve confidently to a wrong answer rather than failing loudly, and every one of them is a call that completes normally while sending responders somewhere other than the caller.

Device location, not user identity

Most multi-site emergency calling failures reduce to one mistaken assumption: that the user's assigned site determines where their calls come from.

It does not, and three ordinary situations prove it:

A visiting employee uses a branch phone. Their extension belongs to headquarters. Their body is in Dallas. The emergency location should follow the handset.

Hot desking. Users sign into different physical phones at different sites.

A shared phone in a workspace. Twelve people use it. It never moves. Its location is a property of the room, not of whoever picked it up.

The rule worth writing on the wall: the phone has the address; the person has the extension.

Why LLDP matters more at a branch than at headquarters

This is the piece missing from most multi-site E911 designs, and it is the piece that makes branch offices harder than headquarters.

Zoom's documentation is direct about the weakness of IP-based location:

"IP addresses are not always helpful — many customers have IP network address translations or subnets that are not sufficiently unique to distinguish a given building or floor of a building from another building or floor."

Branch networks are exactly where that bites. Cookie-cutter branch builds reuse the same private subnet at every site, and centralized internet egress gives every site the same public IP. Both of the IP-based signals collapse, and they collapse silently — the system matches, it is just matching the wrong site.

The signal that survives is switch data, and it carries a prerequisite most multi-site designs never account for:

"Zoom Workplace applications leverage Link Layer Discovery Protocol (LLDP) to determine the network topology in a corporate network based on switch ports."

Specifically, "when making an emergency call, the Zoom Workplace app retrieves the MAC address of the switch from the 'Chassis ID' type-length-value (TLV) and the switch port number or interface ID from the 'PortID' type-length-value (TLV)."

Three consequences for a branch rollout:

1. LLDP has to be enabled on branch switches. If it is off, your most precise signal does not exist, and you fall through to the IP signals that cannot tell your sites apart.
2. The switch inventory is emergency calling infrastructure. Chassis IDs and port IDs have to be recorded and kept current, which makes a network refresh an E911 event.
3. Branch switch replacements change Chassis IDs. Swapping a failed access switch silently unmaps every phone behind it.

Centralized egress and SD-WAN

An SD-WAN or hub-and-spoke design that hauls branch traffic to a central breakout makes every site present the same public IP. If your location model leans on public IP, every branch resolves to the hub.

Do not assume the internet exit point is the physical location of the phone. Where traffic leaves your network is a routing decision. Where the caller is standing is not.

When phones move

A desk phone that moves keeps working. That is the trap.

The last row deserves a decision rather than a discovery. A cloud desk phone will happily register from a spare bedroom and report whatever address it was given at the office.

For any organization with real location complexity, phone moves should raise a ticket. A move request is the natural trigger for the checks that keep the model true: new physical location, new device address, new switch port, remap, retest.

Site lifecycle

Opening a site gets attention. The other three events are where stale emergency data is manufactured.

Opening. Add emergency calling to the branch opening checklist alongside circuits and cabling. The site is not ready because inbound and outbound calling works. Confirm the Zoom Phone site, its emergency address, sublocations, LLDP and switch mappings, desk and common area phones, routing, notification recipients, and the BYOC position, then test before production use.

Moving. Treat it as a new site rather than an edit. Address, sublocations, device addresses, switch and BSSID mappings, IP ranges, notification recipients, carrier records where BYOC applies — then retest.

Merging. The most under-managed of the four. Consolidation leaves debris: phones relocated from the closing site still mapped to it, numbers reassigned, old emergency addresses still live, orphaned network mappings, notification groups pointing at people who moved on. Put emergency calling cleanup in the migration plan, not the punch list.

Closing. Retire the location data. A stale emergency address becomes actively dangerous the moment a number, device or user is reassigned while the old association survives.

Who gets told, locally

Most designs notify a central point. At a multi-site organization that is necessary and insufficient.

The people who can physically help — unlock a door, meet the ambulance, direct responders to the right stairwell, know which entrance is closest to the loading bay — are at the branch. A notification that reaches only corporate security three states away satisfies the letter of the requirement and does very little in the first four minutes.

Decide per site whether response is central, local, or both. Then check the recipient list is a role that still exists. Notifications to a departed branch manager's mailbox are the most common silent failure in this whole design.

The install check

Give this to whoever physically installs a phone at a branch. It is short on purpose.

1. Correct phone, correct assignment, and note where it is physically going.
2. Correct Zoom Phone site, and confirm that site's emergency address is the branch, not an inherited default.
3. Building, floor, suite or sublocation confirmed where the facility needs it.
4. Switch and port recorded; LLDP confirmed enabled on that switch.
5. BSSID or IP mapping recorded where those are used.
6. Emergency routing and carrier architecture confirmed.
7. Internal notification recipients confirmed, including the local one.
8. Validation performed using the approved procedure.
9. Result written down, with an owner's name against it.

Testing across sites

Every site should appear in the validation program. Use Zoom's supported test procedure rather than placing live 911 calls to experiment — the controlled method using an internal safety response team and maximum holding time is described in the configuration guide.

Cover at minimum: a desk phone at each site type, a common area phone, a phone on a non-primary floor, a hot desk if you use it, a recently moved phone, and a BYOC site if you have one. Write down the expected location before you test. A test that only confirms a call connected has confirmed nothing about location.

Zoom's Emergency Services dashboard reports which IP phones map to known locations, which is the standing report for this. For a multi-site estate it is the fastest way to see a branch that was built without complete location configuration.

Frequently asked questions

Does each Zoom Phone site need its own emergency calling configuration?

Yes. Zoom's documentation states that when multiple sites are enabled, emergency calling must be set up for each site. Creating a site does not inherit a correct emergency configuration.

Can a branch use headquarters as its emergency address?

Not if it does not describe where the caller is. A headquarters address on a branch phone is the specific failure this article exists to prevent.

Does the user's extension determine the 911 address?

No. Emergency location depends on the device, the site, detected network data, the carrier architecture and the configuration. Assigning an extension is not assigning a location.

What happens if a desk phone is moved to another building?

If the destination network is correctly mapped, detection may follow it. If not, the phone can keep reporting its old location indefinitely while working normally. Moves belong in change management.

Why does LLDP matter for branch offices?

Because branch networks commonly reuse subnets and share a central internet breakout, which makes IP-based location unable to tell sites apart. Switch data via LLDP is the signal that still distinguishes them, and it only works if LLDP is enabled.

Does SD-WAN affect emergency location?

It can, and this is our assessment rather than documented Zoom behavior. Centralized egress makes every site present the same public IP. Do not treat the internet exit point as the physical location.

What about hot desking?

Zoom does not document hot desking behavior for emergency calling. Test it explicitly. The physical phone's location should take precedence over the signed-in user's home site.

Does BYOC change what the branch reports?

Yes. When a BYOC carrier handles the emergency call, the address delivered to the PSAP comes from that carrier's records rather than Zoom's detected location. See the configuration guide.

Related articles

• Does Kari's Law apply to my phone system? — what the rules require
• How do you configure Zoom Phone for Kari's Law and RAY BAUM's Act? — the build, detection mechanics, BYOC and ELIN
• What counts as a dispatchable location for remote and hybrid workers? — people, not places
• Zoom topic hub — every live Zoom Phone, E911 and Zoom administration article in one place.

References

1. FCC, Multi-line Telephone Systems: Kari's Law and RAY BAUM's Act 911 requirements
2. FCC, Report and Order 19-76 — PS Docket Nos. 18-261 and 17-239, GN Docket No. 11-117
3. Zoom, Deploying a nomadic emergency services solution
4. Zoom, Monitoring emergency services readiness with Zoom Phone
5. Zoom, Setting up nomadic emergency services (KB0062169)
6. Zoom, Setting up emergency response locations, ERL and ELIN (KB0064642)

ADAM Pulse is USA Telecom's managed Zoom Phone and network practice. We map sites, switch ports and emergency addresses so a branch desk phone reports the building it is in, not headquarters.