UCaaS day-two operations
Go-live is the switch-on; day two is everything after — a number that stops answering at 6 PM, a new hire who needs a phone, a port that finally completes, a destination that will not place. This runbook collects the seven day-two operations an SMB operator performs on the phone system into one ordered page, so each recurring task lands on the right guide instead of a search across ~80 voice pages. Every control below is yours — the tenant’s. The platform supplies the surface; the wiring, the assignment, the country list, and the failover choice remain your configuration.1. Route live traffic: failover numbers
The one setting that decides whether a problem is an outage or a shrug: the per-DID fallback route. A live number without a fallback has one fragile destination between the caller and a dropped call. Open the Numbers routing page on each published DID and confirm:- Primary route — the queue, ring group, IVR, SIP forward, or voicemail box the DID normally answers with.
- Fallback route — set
fallback_type+fallback_configto the destination that must answer when the primary fails (queue has no one available, webhook times out). On an SMB phone system the correct fallback is almost always voicemail: an empty roster or an unregistered phone then takes a message instead of dropping the caller. - Business-hours gate — if the DID should behave differently outside staffed hours, put a schedule gate in front of the route rather than un-staffing it.
2. Country coverage: outbound vs inbound eligibility
Two different questions, often conflated in day-two troubleshooting:- Can I place outbound calls to this destination? Read the per-country verdict on Voice → Country coverage. The page resolves each destination to an enabled / restricted / disabled verdict — an agent whose outbound list crosses into a restricted country fails the dial with a coverage answer, not a carrier error. When a specific destination must stay reachable, gate the agent’s dial list with the tenant-owned country allowlist so the choice is explicit configuration, not accident.
- Can I receive inbound calls / SMS on a number of this country? That is a number-inventory question: buy or port a number in the destination country (or one the destination can reach), then route it. Coverage constraints live at the tenant level; the matrix tells you the verdict, your allowlist owns the choice.
3. Choose a distribution primitive: the five-primitive decision tree
“Put the sales line on a hunt group”, “parking this for accounts”, “page the warehouse”, “agents share desks” — each maps to exactly one of the five Voice primitives, and picking the wrong one is the most common day-two misconfiguration. The shortcut:- Everyone with a desk rings at once, anyone answers → Shared Line Appearance.
- Fan out the inbound call across a team by strategy (ring-all, sequential, round-robin) → Ring groups.
- Move a live call to a colleague, retrieved from any phone → Call park.
- One-way announcement to desk speakers → Paging groups.
- More agents than desks; calls follow the signed-in person → Hot-desking.
4. Provision endpoints: SIP extensions, and optionally your own PBX
New hire, new desk, or a handset replacement — the task is always: create a SIP extension, point a softphone or desk phone at it, confirm it registered.- Extensions. Create SIP credentials per person under Voice → Extensions; the same credential feeds the browser softphone, a desktop SIP app, or a physical handset. Polycom, Yealink, Cisco, and Grandstream handsets pull their config over zero-touch provisioning, and a CSV bulk import rolls out a whole office at once. Walkthrough: Extensions and desk phones.
- Your own PBX. If the estate already runs Asterisk, FreeSWITCH, or 3CX, register the trunk on Orbit as the outgoing side and keep your on-prem dial plan — reachable numbers then ride the existing PBX. Host/port/transport setup, register strings, NAT and auth failure diagnosis, and the registration health dashboard: Register your PBX on Orbit.
- Watch the health. A phone that was provisioned but never registered answers “not online” in the extension health list; check registration before checking routing — the routing page cannot fix an endpoint that has not signed in.
5. Porting and assignments: number lifecycle
A DID moves through two recurring day-two operations — getting it onto the platform, and assigning it to the team that owns it.- Port in. Preflight the number, gate on eligibility, submit the LOA, checkpoint wizard drafts when documents are incomplete, ride the FOC date, and move at bulk scale with one CSV (up to 1000 DIDs). Procedure end-to-end: Port a number; batch mechanics: Port numbers in bulk.
- Assign. Decide which user, queue, ring group, or DNI tracking pool owns each number — from the per-number detail page or the bulk Assignment Portal. A port finishing with no assignment plan leaves the new DID answering on the default route. Procedure: Assign and reassign numbers.
- Port out / caveat. A departing number unwinds its assignments first; the checklist that separates a clean port-out from a dropped call is in Port out numbers.
6. Inbound routing hints: per-DID composition
After §3 picked the primitive and §5 assigned the number, compose them on the DID:- One DID = one primary route + one fallback route (§1) — point the primary at the ring group, shared line, queue, or SIP forward you chose.
- DNIS pattern routing — batch a block of DIDs (e.g. a whole ported range) onto one route by matching the dialed-number pattern instead of editing each DID.
- Ring rules — the per-user answer rules (which devices ring, in what order) sit underneath the primitive: the shared line rings devices, the ring rule decides which of a member’s devices.
- Verify the whole composition with the IVR simulator before you publish the route — a simulated call answers in seconds, a misrouted production number answers never.
7. When IVRs and queues enter the picture
Day-two scope creeps toward the contact center in two ways; catch them early:- IVR — add a menu in front of the distribution primitive when callers in the same pool need to self-select (press 1 for sales, 2 for support). The IVR is an interactive front door on the DID; the primitives above keep doing the actual fan-out.
- Queues — graduate from a ring group or shared line to an ACD queue the moment callers must FIFO-hold, agents must be ranked by skill, or an SLA timer needs tracking. The queue model, SLA timers, and overflow actions: The ACD queue model; the setup and run procedure: Voice queues.
Day-two checklist
- Weekly: audit DID fallbacks; fill unset ones (§1).
- On every outbound-failure report: check the coverage verdict before the carrier (§2).
- On every “we need a hunt group” request: route through the five-primitive tree (§3).
- On every hire/desk change: provision the extension; check registration health first on any “not reachable” report (§4).
- On every port completing: assign the DID in the same session (§5).
- Before publishing any route change: verify with the IVR simulator (§6).
- Quarterly: review whether a busy ring group has graduated into FIFO-hold territory (§7).