Queues

Where an inbound caller waits, who is offered the call, and what today’s queue actually does with the strategy you chose.

A queue holds callers and offers them to agents. It lives under Inbound, beside processes and numbers, because a queue is a property of how a number is answered — not of a shift, and not of the phone system’s inventory.

What happens to a caller, in order

1. Answered, then music

   The call is answered and the hold music starts at that moment, before anybody’s phone rings. Resolving who to ring takes a moment, and silence in that moment is what a caller hears as a dead line.

2. Every member's every device rings at once

   Not one device per person: an agent who takes queue calls in the browser as well as on a handset has both ring, exactly as a direct call to them does.

3. The first to answer wins

   Everybody else stops ringing, and the call is attributed to whoever actually picked it up.

4. Or the wait runs out

   After Max wait the call leaves the queue. If the queue takes messages, the caller is offered voicemail; otherwise they are told nobody is free and the call ends.

The strategy is stored, and today every queue rings everyone. The four choices — ring everyone, longest idle first, take turns, in order — are saved on the queue and shown back, and the call path is a ring-all against the available members whichever is chosen. Full distribution, with skills and wrap-up, is still to come. Choose the one you want; expect ring-all until then.

Members

Members are agents, rostered individually, each with a priority.

A member is rung when they are a human agent whose extension is enabled. Priority orders the list and, because the ring is simultaneous, does not yet stagger who is tried first.

A virtual agent can be added to a queue and is not rung by it. The queue path asks for human members only. To have a virtual agent answer, point the number or the flow at it directly — see virtual agents.

Rostering needs queues:write, which a supervisor holds as well as an admin. Adding somebody who is already a member replaces their priority rather than failing.

Settings

Setting
Hold music	this queue’s own, otherwise the organisation’s, otherwise the switch’s
Strategy	see the note above
Max wait	seconds before the call leaves the queue. 0 is no limit; a queue that has never been given one behaves as 60 seconds
Team	which team this queue belongs to, for filtering and reporting
Record calls	see below
Voicemail	whether a caller who could not be connected is offered a message

Hold music is resolved most specific first, so a queue with its own music costs nothing extra to play — see the audio library.

When nobody can be reached

Two different failures, deliberately told apart:

What the caller hears	Recorded as
nobody was available, or a voicemail greeting	failed, or voicemail — the queue had nobody to offer
the wait ran out while they were on hold	timed out, or voicemail
they hung up first	abandoned

Those demand opposite responses from whoever reads the report: staff the queue, shorten the wait, or find out why callers are giving up. A timeout and an abandon are told apart by the clock rather than by what the switch called the error, because that text would silently reclassify every one of them the day it changed.

The stock “nobody is available” wording is the closest sentence the switch ships. A tenant with a real queue records their own.

What you see live

Waiting calls and the longest current wait, from the live wait tracker rather than from the call records, so every node agrees about the same queue.

Where the figure cannot be read, FireTone reports it as unavailable rather than as zero. A staffing decision made on an invented number is worse than one made on an honest “unknown”.

Each visit to a queue is its own record, so a call sent from one queue to another keeps both. A single queue against the call would keep only the last one, losing the queue that actually made the caller wait.

Recording

Calls answered through this queue can be recorded by switching Record calls on here.

Recording is on if any level says so — organisation, team, extension, queue, trunk or process — so this can add recording and cannot switch off a policy set above it. See Recording, and what is kept.