What Is a Channel in Telephony?

What a channel is

In the days of old phone lines, the answer was simple: one line = one call. An office with three lines could handle three outside calls at once, and the fourth caller heard a busy tone. The word channel was born to describe exactly that — one "slot" for one call on the organization's connection to the outside world.

Today, when calls travel over the internet through a single connection called a SIP trunk, there is no longer a wire for each call. But the concept remains: a channel is one outside call happening right now. If you have eight channels, eight calls can take place at the same time, whether incoming or outgoing.

The diagram shows the principle: the PBX has 40 extensions but four channels. Four callers get in, and the fifth hears a busy tone — even if 36 employees are sitting free next to their phones.

Extensions vs. channels — the common confusion

An extension is an endpoint: an employee's phone, app or mobile. A channel is capacity for outside calls. These are two separate things. An organization can have 50 extensions, and at a given moment only 8 of them are on an outside call.

• An internal call between two extensions doesn't use an outside channel — it never leaves the PBX.
• An incoming call waiting in a queue does use a channel, even if no one has answered it yet. The caller is connected to the PBX and hears music.
• Transfer out — a call that came in and is transferred to an employee's mobile outside the PBX usually uses two channels: one for the caller, one for the outgoing leg to the mobile.
• A conference with five outside participants uses five channels.

The second point is especially important: a waiting queue does not save channels. It saves agents. Anyone waiting in a queue holds a channel for the whole wait.

How to estimate how many channels you need

There is a simple method that gives a good direction. Take the busiest hour of the day and ask two questions: how many calls take place in that hour, and how long does a call last on average (including the time waiting in the queue).

The formula: number of calls in the peak hour × average call length in minutes ÷ 60 = the average number of calls happening at the same time. For example: 120 calls an hour, 4 minutes each: 120 × 4 ÷ 60 = 8 simultaneous calls on average.

But calls don't arrive in an even stream — they arrive in waves. So you add a margin: in small organizations it is common to add a few channels above the average, and large call centers use a more precise statistical calculation called Erlang — named after a Danish engineer who developed traffic theory in telephony in the early twentieth century. The unit of telephone traffic is still called an "erlang" in his honor.

Where do you get the numbers? From the call log. On an existing PBX you can see exactly how many calls there were in each hour and how long each one lasted. If you are moving from an old system, you can ask the previous provider for the data, or estimate by the number of employees who answer the phone in a busy hour.

Campaign night: when the peak is ten times normal

A nonprofit that gets 30 calls on a regular day discovers on a concentrated fundraising night that the phone never stops. A radio host mentions the number, a message goes out in groups, and within minutes hundreds of people dial at once. This is the moment when the number of channels stops being a technical detail and becomes a question of money — every caller who heard a busy tone is a donation that may never come back.

• Plan for the peak, not the routine. The capacity of a regular day won't hold up on a night like this.
• A queue instead of a busy tone. A waiting queue takes in the callers, plays them a message and music, and connects each one to an agent as one becomes free.
• More hands. Temporary extensions for volunteers — from home, from a mobile or from the call center — increase the answering rate, and so the wait gets shorter.
• Callback. Anyone who doesn't want to wait can leave a number, and the system will call them back when an agent becomes free — which shortens the queue.

The same principle holds for a business in Bnei Brak before the holidays, for a yeshiva on the evening when parents call to ask about admission, and for a service center on a day when there was a general outage.

What happens when the channels run out

When all the channels are busy, the next caller doesn't enter the PBX at all. They hear a busy tone or a message from the network, and the PBX doesn't "know" they tried — so in many cases that call won't even appear in the call log. This is why organizations sometimes can't understand why customers complain that "you can't be reached" while the agents see an empty queue.

Outgoing calls get blocked too: an employee who tries to dial out when all the channels are busy will get a busy tone. In an organization that runs an outbound dialing campaign and also receives calls, it is worth thinking ahead about how the two will live together at the same hour.

Practical tips

• Check the real peak hour in the call log — not the daily average.
• Remember that waiting in a queue, transfers to mobile and conferences all use channels.
• For a one-time event, talk to the provider a few days ahead, not on the day of the event.
• Also check the bandwidth in the office — eight simultaneous calls on a weak internet connection will sound bad even if you have the channels.
• After the event, look in the log at how many calls there really were at the peak. That is the basis for planning next time.

Three worked examples: an office, a yeshiva, a nonprofit

The formula from the previous section sounds simple, but the difference between organizations is in the question of what you count. Here are three full calculations, with the considerations behind them.

1. A law office, six employees. In the call log of a busy month, you find that the peak hour is between ten and eleven in the morning: 24 incoming calls and 12 outgoing, 36 in total, with an average length of three and a half minutes. The calculation: 36 × 3.5 ÷ 60 = 2.1 simultaneous calls on average. But calls arrive in waves — when the court clerk calls, two lawyers dial clients at the same moment. So you double it and add one: 4 channels. Fewer than that, and once a week someone will hear a busy tone. More than that, and it's money spent on channels that go unused. Notice that the number of employees, six, doesn't enter the calculation at all.

2. A yeshiva with a dormitory, 180 students. Here there are two separate streams. The first stream is the office: the secretaries, parents who call, suppliers — in the peak hour about 15 calls of four minutes, meaning one simultaneous call on average, three channels with a margin. The second stream is the dormitory phone stations: 12 phones in the hallways, all of which come alive in the one-hour window after the evening study session. If 120 students call home in that hour, six minutes each, the calculation gives 120 × 6 ÷ 60 = 12 simultaneous calls — exactly the number of phones. That means that here the queue doesn't form on the phone but in the hallway: whoever has no free phone waits on the side. The conclusion: 15 channels in the evening (12 for the stations, 3 for the office), and during the day 3–4 are enough. And another important lesson: if the dormitory stations and the office share the same channels, then on an evening when all 12 stations are in use, the secretaries won't be able to dial out — so the margin for the office is not a luxury. What really moves the calculation is the number of stations, the length of the calling window, and the calling cards that limit each student's call length.

The third example: a nonprofit on campaign night, when the agents are the bottleneck

3. A nonprofit on a fundraising night. On a regular day the nonprofit gets 30 calls, and two channels would be enough for it. On campaign night it expects 3,000 calls in three hours, and in the peak hour — after the radio host mentions the number — 1,500 calls an hour. A donation call lasts two and a half minutes, plus an average of a minute and a half waiting in the queue: four minutes during which the caller holds a channel. The calculation: 1,500 × 4 ÷ 60 = 100 simultaneous calls on average, and in the waves — 150 or more.

But here something shows up that the formula doesn't reveal. That evening 40 volunteers are sitting at the phones. 40 volunteers, each of whose calls lasts two and a half minutes, together finish 40 × 60 ÷ 2.5 = 960 calls an hour. 1,500 are coming in. That means that even with a thousand channels the queue would only get longer — every minute more calls come in than go out. The channels determine how many people wait; the agents determine how many get answered.

• More hands — to handle 1,500 calls an hour you need about 63 volunteers, or a shorter call. Temporary extensions for more volunteers working from home solve this without any equipment.
• Callback — anyone who chooses callback frees the channel immediately, and the system calls them when a volunteer becomes free. This way the same 100 channels serve many more people.
• A short message — a 40-second greeting in the menu before the queue is 40 seconds of a busy channel for no benefit. On a night like this, shorten it to 10.
• Volunteers as extensions, not as transfers — a volunteer who receives calls transferred to their mobile through the telephony network uses two channels per call. A volunteer registered as an extension (an app or an IP phone) uses one.
• Outgoing calls afterward — thank-you or reminder calls on the dialer should be scheduled for the next day, not for the peak hour, because they share the same channels.

The summary for the nonprofit: order about 150 channels for the evening, recruit agents according to the call rate and not by gut feeling, and turn on callback. All three together — not just one of them.

When the channels run out: how to spot it in real time, and what to do at that moment

The previous section explained what the caller experiences. The practical question is how you find out — because from the inside everything looks normal. The order of events when capacity fills up is almost always the same:

1. Outgoing calls are hit first. An employee picks up the handset, dials, and gets a busy tone or an error tone. This is the first complaint to arrive, and it is usually not connected to the incoming load.
2. The queue stops growing. On the queues screen, the number of waiting callers gets stuck at a fixed value, even though from the outside "everyone is calling". A queue that doesn't grow during the peak is a warning sign, not a good sign.
3. New callers aren't recorded. In the call log for that hour you see a flat ceiling: the number of simultaneous calls reaches a certain value and doesn't go above it, hour after hour. That ceiling is your number of channels.

What you can do within the minutes of the event, without waiting for the provider: turn on callback in the queue, shorten the message before the queue, pause an outbound campaign running in parallel, and move volunteers who work through transfer to mobile over to working from an app. After the event, open the log and look at the ceiling — that is the number to ask the provider for next time, plus twenty percent.

The first four can each run out separately, and each looks different from the outside. Before buying channels, make sure that is really what ran out.