What is a VoIP phone system, how it works and what you need
A VoIP phone system handles your business calls over the internet instead of traditional phone lines. This guide explains what it is, how a call travels from caller to agent, what you need to run one and the mistakes to avoid when you pick a system.
What is a VoIP phone system?
A business phone system, often called a PBX (Private Branch Exchange), is what receives the calls to your company and decides where they go: to a person, a department, a voice menu, voicemail. It works the other way too, letting staff call out using the company number and reach each other with a short extension number.
A VoIP phone system does the same job, but carries voice as data over an IP network, the same network your email and web traffic use. VoIP stands for Voice over Internet Protocol. Instead of a copper pair for every line and a cabinet full of telephony cards, you have software that talks to desk phones, browsers and your carrier over your internet connection.
The system can live in your office, on a server or a dedicated appliance, or in the cloud, run by a provider and managed from a web portal. The underlying technology is the same; what changes is who maintains it, how you pay and where people can connect from. If you are weighing up the two, see our comparison of cloud and on-premise PBX.
How VoIP works: SIP, RTP and codecs
Every VoIP call has two separate parts. Signalling covers who is calling whom, whether the phone is ringing, whether it was answered and when the call ends. Nearly every phone system and carrier handles this with SIP (Session Initiation Protocol). The audio itself travels separately, in small packets carried by RTP, usually one every 20 milliseconds.
Before it leaves, your voice is digitised and compressed by a codec. The common ones are G.711 (PCMU or "u-law" in North America, PCMA or "a-law" in Europe), which matches traditional phone quality without compression; G.722, a wideband codec that sounds more natural; and Opus, widely used for browser calls because it adapts to network conditions. Both ends of a call need at least one codec in common, or something in the middle has to transcode.
When you press a key during a call, say to pick an option in a menu, that tone has to reach the system reliably. It is called DTMF and can travel inside the audio (in-band) or as a separate event (RFC 2833, now RFC 4733). If the phone system and the carrier disagree on the method, key presses are not recognised, which is one of the most common problems in a new setup.
- SIP sets up, manages and ends calls, and registers phones with the system.
- RTP carries the audio packets; its encrypted form is SRTP.
- Codecs compress and decompress voice (G.711, G.722, Opus and others).
- DTMF is the key presses during a call, sent in-band or as events.
- A SIP trunk connects your phone system to the carrier that provides your numbers and reaches the public phone network.
The building blocks of a VoIP phone system
It helps to take a VoIP system apart into its pieces. Each has a clear role and, in a well-designed system, you can swap one without touching the others: change carrier without replacing phones, or replace phones without reconfiguring your numbers.
| Component | What it does | Examples |
|---|---|---|
| Phone number or SIP line | The number customers dial, supplied by a carrier and connected to the system | Local number, mobile number, toll-free number |
| Phone system (PBX) | Receives calls and routes them according to your rules | On-site software or a cloud service |
| Extension | One person’s SIP account, with a short number | 201, 301, 302 |
| Device | What the agent speaks and listens on | IP desk phone, softphone, browser webphone |
| Routing modules | The rules that decide a call’s path | IVR menus, queues, business hours, transfers |
| Outbound routes | Which line an outgoing call uses and which caller ID it shows | Mobiles via one line, international via another |
How an inbound call flows, step by step
Here is what happens when a customer dials your business number. The example follows the logic of a cloud phone system like Robottino, but the sequence is similar in most systems.
The call reaches the carrier
The customer dials from a mobile or landline. The public phone network hands the call to the carrier that hosts your number, which forwards it to your phone system over SIP.
The system identifies the dialled number
Each number has its own rules. Blocked callers are rejected first, then the system checks whether the number has hit its limit of simultaneous calls.
Business hours check
If the number is linked to a schedule, the system checks whether you are open right now. Out of hours, the caller hears a message and the call takes a different route.
Voice menu or AI agent
During opening hours the call may go through a "press 1 for…" menu, or an AI voice agent that understands the request in plain speech.
Queue and agent
If everyone in the team is busy, the call joins a queue with hold music and announcements. As soon as someone is free, their phone rings and the conversation starts.
Logs and reports
Afterwards everything is on record: each step with its timestamp, who answered, how long the caller waited and the recording, if there is one.
Key VoIP phone system features
The basics, like transfer, hold and voicemail, are a given today. What sets a modern VoIP phone system apart is how well it gets every call to the right person at the right time without losing context along the way.
- IVR voice menus: callers pick a department on the keypad; with an IVR menu you type the prompts and a synthetic voice reads them.
- Call queues: callers wait in line instead of getting a busy tone, and the call goes to the next free agent according to a strategy.
- Business hours and holidays: calls take different routes when you are open, closed or away.
- Transfers to outside numbers, such as an on-call engineer’s mobile.
- Caller recognition from a shared phonebook, on the phone and in the agent panel.
- Call recording with a time limit on how long recordings are kept.
- Reports and real-time monitoring: missed calls, wait times, agent workload.
- Automatic callbacks and AI voice agents so nobody is left unanswered.
- CRM and system integrations through webhooks and APIs.
What you need: network, bandwidth and devices
Voice copes badly with delay and lost packets, far worse than a web page does. That is why connection quality matters more than headline speed. With G.711, a call uses roughly 80 to 100 kbps in each direction including network overhead; compressed codecs use much less. An ordinary fibre line can carry dozens of simultaneous calls, but only if it is not saturated and the router does not queue voice behind a big download.
Three measurements matter. Latency, the delay: ITU-T guidance is to stay under 150 ms one way for natural conversation. Jitter, the variation in that delay, which makes audio choppy. And packet loss, which you hear as clipped words. A wired connection is almost always more stable than Wi-Fi.
For devices you have three options, and you can mix them. An IP desk phone plugs into the network like a computer. A softphone is an app on a computer or smartphone. A webphone runs straight in the browser using WebRTC, with nothing to install. We cover them in detail in our guide to calling from the browser.
- A stable internet connection, ideally with voice traffic prioritised (QoS) on the router.
- A headset with a microphone for anyone working at a computer; it cuts echo and background noise compared with speakers.
- One or more numbers from a SIP carrier, either new or ported from your old provider.
- A firewall that allows SIP and RTP, and a router with SIP ALG turned off, since it often mangles SIP packets.
Pros and cons of a VoIP phone system
The biggest advantage is flexibility. An extension is no longer tied to a socket in the wall: an agent can answer from the office, from home or from another site with the same number and the same rules. Adding a person means creating an account, not running a cable. Changing how calls are routed happens in a web panel, often in minutes.
The second advantage is integration. A VoIP call is a digital event, so it can be recorded, matched to a customer record, counted in reports and pushed to a CRM. That is what makes features like caller recognition or conversation analysis possible.
There are limits worth knowing. Without internet, office phones cannot take calls, so plan a fallback such as forwarding to a mobile. Audio quality depends on the network, and an office on a shaky connection will notice. Finally, check with your carrier how emergency calls are handled and what location they report, especially for remote workers.
How to choose a VoIP phone system, and mistakes to avoid
The right question is not how many features a system has but how easily it lets you handle calls the way you actually work. A list of a hundred features is no help if changing your opening hours needs a technician.
- Can you use numbers from any SIP carrier, or are you tied to the vendor’s own?
- Can you change hours, menus and queues yourself from a portal, with immediate effect?
- Can agents work from a browser, a softphone and a desk phone, including away from the office?
- Do reports show missed calls, wait times and the full path of each call?
- Do recordings stay private, with a retention period you control?
- Is pricing clear: what is monthly, what is pay-as-you-go and what is included?
- Does it integrate with your CRM or other systems through webhooks or an API?
Example: a VoIP setup for a small business
Picture a practice with one main number, three people in accounts and two in technical support. A simple, effective setup: the number leads to a menu with two options, accounts and support. Each option feeds a queue with its own agents. Out of hours, a message gives the opening times and offers to take a number for a callback.
In a cloud phone system like Robottino you build this from the portal: connect your carrier’s number under phone numbers, create the voice menu and the call queues, and agents answer from the agent panel in the browser, or from a softphone or IP phone. For each piece in more depth, see our guides on designing an IVR menu, call queue strategies and business hours and holidays.
As call volume grows, the same pattern scales: an AI voice agent can deal with common questions before the queue, and an automatic callback stops customers hanging up after a long wait. Those steps are covered in our guide on how to reduce missed calls.
Frequently asked questions
What is the difference between a VoIP phone system and a traditional PBX?
A traditional PBX uses analogue or ISDN lines and dedicated hardware cards. A VoIP phone system carries voice as data over the internet using SIP, so extensions can be anywhere with a connection and the routing rules are changed in software.
Can I keep my phone number when I move to VoIP?
In most cases yes, by porting the number to a SIP carrier. Check timescales and paperwork with the new carrier, and start the port only once the new system is configured and tested.
How much bandwidth does VoIP need?
It depends on the codec and the number of simultaneous calls. With G.711 a call uses about 80 to 100 kbps each way; compressed codecs need less. Stability matters more than speed: low latency, little jitter and no packet loss.
Do I need special phones for VoIP?
No. You can use an IP desk phone, a softphone on a computer or smartphone, or a webphone that runs in the browser. Old analogue handsets can be connected through an ATA adapter.
What happens if the internet goes down?
Office phones cannot take calls until the connection is back. With a cloud phone system the calls still reach the system, so you can route them to a mobile, to colleagues at another site or to a message.