VoIP Bandwidth Requirements for Australian Businesses: Getting the Numbers Right

Setting up a cloud phone system or troubleshooting poor call quality usually leads to the same question: how much internet do you actually need for VoIP? The honest answer is that raw bandwidth is rarely the problem. Most businesses have enough download speed. What trips them up is latency, upload capacity, and traffic prioritisation — and those are different problems with different solutions.

This article walks through the real numbers: how much bandwidth each VoIP call consumes, how to calculate your total requirement, and why your internet connection might be failing your phone system even when your speed test looks fine.

Why VoIP Is More Demanding Than You Think

Most internet applications are tolerant of imperfect connections. A web page loads a fraction of a second slower. A file download takes an extra few seconds. A video stream buffers briefly and catches up. These applications are asynchronous — they request data, receive it, and reassemble it at their own pace. Timing imperfections are invisible to the end user.

VoIP is different. A phone call is a real-time, two-way audio stream. There is no buffer, no catch-up mechanism, and no second chance. When a packet of voice data is delayed, it does not arrive late — it arrives too late to be used, and the audio drops out. When packets arrive out of order or at irregular intervals, the audio becomes choppy. When the round-trip delay between speakers exceeds a threshold, conversation becomes awkward and difficult to follow.

This is why an NBN connection that handles Netflix, Zoom, and cloud backups without complaint can still deliver frustrating VoIP call quality. The speed test says 80 Mbps down. The calls still sound broken. Speed is only one dimension of internet quality, and for VoIP it is not even the most important one.

Three metrics determine whether your internet connection will support VoIP reliably: bandwidth, latency, and jitter. All three need to be within acceptable ranges simultaneously.

The Three Metrics That Determine VoIP Quality

Bandwidth

Bandwidth is the raw capacity of your internet connection — the amount of data that can be transmitted per second. For VoIP, what matters is not just your total bandwidth but your usable bandwidth during peak periods, and specifically your upload capacity, since voice data leaving your office travels via the upload path.

Each VoIP call consumes a fixed amount of bandwidth in both directions for the duration of the call. The amount depends on which audio codec is in use. That bandwidth requirement is constant and non-negotiable — unlike a web request, VoIP cannot simply slow down and try again.

Latency

Latency is the time it takes for a data packet to travel from your device to the remote server and back — the round-trip time. For VoIP, what matters is one-way latency: the time for your voice to travel from your phone to the other party's phone.

The ITU-T G.114 standard recommends that one-way latency remain below 150 milliseconds for acceptable voice quality. Below 50ms is ideal. When latency exceeds 150ms, speakers begin to perceive a noticeable delay. At 250ms and above, the delay is prominent enough to cause the familiar "talking over each other" effect, where both parties begin speaking simultaneously because neither can tell the other has finished.

Latency is determined by physical distance, network path, and the type of connection technology. A business on NBN FTTN (fibre to the node) may see higher latency than one on FTTP (fibre to the premises) simply due to the copper segment between the node and the premises. This is a structural characteristic of the connection, not something that can be resolved by upgrading to a faster speed tier.

Jitter

Jitter is the variation in packet arrival times. Even if the average latency of your connection is acceptable, if individual packets are arriving at wildly different intervals — some taking 20ms, others 80ms — the receiving device cannot reconstruct a smooth audio stream.

For acceptable VoIP quality, jitter should remain below 30 milliseconds. VoIP phones and software clients use a jitter buffer — a small reservoir of incoming packets that smooths out the variation before playback. The jitter buffer is effective up to a point, but it introduces a small additional latency of its own. When jitter is severe, the buffer cannot compensate without adding unacceptable delay, and the audio quality degrades.

Jitter is often caused by network congestion — packets from different applications competing for the same bandwidth and being delayed unevenly. This is particularly relevant during business hours when internet usage peaks, which is why VoIP problems that only appear during the workday are frequently a jitter problem rather than a bandwidth problem.

VoIP Bandwidth Per Call — The Numbers

The bandwidth consumed by a VoIP call is determined by the codec in use. A codec is the algorithm that compresses and encodes voice audio for transmission. Different codecs make different trade-offs between audio quality and bandwidth consumption.

The bandwidth figures above include IP, UDP, and RTP packet headers in addition to the voice payload. The actual audio data is a smaller component; the overhead from packet headers adds significantly to the total, particularly for smaller packet sizes.

G.711 is the most widely deployed codec in Australian cloud phone systems and is equivalent in audio quality to a traditional PSTN call. G.722 uses the same bandwidth but captures a wider frequency range, producing noticeably clearer and more natural-sounding voice — the difference is audible on any decent handset or headset. G.729 achieves significant compression at the cost of audio quality, and is generally reserved for constrained environments like satellite internet or remote mobile connections where bandwidth is limited and expensive.

For planning purposes, use a practical rule of thumb: allow 100 Kbps per concurrent call. This 100 Kbps figure provides a buffer above the codec's theoretical requirement, accounting for overhead, retransmission, and minor variations in packet size. It is a conservative estimate that holds across all common codecs and simplifies the planning calculation.

How to Calculate Your VoIP Bandwidth Requirement

The calculation is straightforward once you understand the input you need: concurrent calls, not total staff.

Step 1 — Identify your peak concurrent call count

This is the maximum number of simultaneous active calls your business places or receives at its busiest moment. For a 20-person office, this is not 20 — it is the realistic maximum of calls happening at the same time. For most businesses, peak concurrent calls run at roughly 30–50% of total staff. A 20-person office typically peaks at 6–10 concurrent calls.

If you have call reporting from an existing phone system, use it. The data will show your peak concurrent call count with precision. If you are setting up a new system, use your best estimate of how many people are on calls simultaneously at the busiest time of the working day.

Step 2 — Multiply by 100 Kbps

Multiply your peak concurrent call count by 100 Kbps. This gives your VoIP bandwidth requirement.

Step 3 — Add to your other internet requirements

VoIP bandwidth is small relative to other business internet usage. A 20-person office with 10 concurrent VoIP calls needs 1 Mbps for the phone system. The same office running cloud applications, receiving email, and performing background updates will consume significantly more. For sizing an internet connection, VoIP is almost never the constraining factor.

The exception is upload capacity, which is covered in detail in the section below.

For broader guidance on total bandwidth requirements, see business internet speed requirements.

Latency — The Real VoIP Killer

The bandwidth calculation above makes VoIP sound easy, and from a download capacity perspective, it is. What actually causes call quality problems in Australian businesses is almost always latency or jitter, not throughput. Understanding the latency characteristics of different connection types helps diagnose and prevent problems.

NBN FTTP (Fibre to the Premises)

FTTP is the best-performing NBN technology for VoIP. The connection runs optical fibre from the exchange directly to the premises, with no copper segment. Typical one-way latency on an FTTP connection is 5–15ms, well below the 50ms ideal threshold. FTTP is the recommended connection type for businesses running cloud phone systems or other real-time applications.

NBN FTTB and FTTC (Fibre to the Building / Fibre to the Curb)

Both technologies run fibre close to the premises with a short copper segment for the final connection. Typical latency is 10–25ms — still well within acceptable thresholds. For most businesses on FTTB or FTTC, latency is not a VoIP risk.

NBN FTTN (Fibre to the Node)

FTTN uses a longer copper segment between the street-level node and the premises. Latency on FTTN connections varies more widely and can run 15–40ms at the NBN layer, with additional latency added by the internet service provider's network. More importantly, FTTN connections are more susceptible to noise and interference on the copper segment, which increases packet loss and jitter — both directly harmful to VoIP quality. If your business is on FTTN and experiencing call quality issues, the connection technology itself may be a contributing factor. Upgrading to NBN FTTB, FTTC, or FTTP where available will address this structurally. See NBN FTTP, FTTB, FTTN and HFC explained for a full breakdown of technology differences.

Fixed Wireless (Commercial Grade)

Commercial-grade fixed wireless internet — point-to-point radio links used in business-grade deployments — typically delivers one-way latency of 5–20ms with very low jitter. It is well-suited to VoIP. Consumer-grade fixed wireless broadband (NBN Fixed Wireless) is more variable and may exhibit higher jitter during busy periods.

Starlink

Starlink low-earth orbit satellite internet has transformed satellite connectivity for regional and remote Australian businesses. Typical latency on Starlink is 20–60ms — significantly lower than traditional geostationary satellite (which runs at 600ms or higher). Starlink is generally usable for VoIP, though the higher end of its latency range approaches the region where call quality becomes noticeable. Jitter on Starlink can also be higher than fixed-line technologies, particularly during periods of satellite handover. It is an acceptable option for remote locations where alternatives are limited, but it is not recommended as a primary VoIP connection for businesses with other options.

4G Failover

A 4G connection used as a failover path during an internet outage will typically deliver 30–80ms one-way latency — adequate for VoIP calls. Latency will be noticeable compared to a normal fixed-line connection but calls will generally be comprehensible. The primary concern with 4G failover is jitter, which can be higher and more variable than fixed-line connections, particularly in congested mobile cells during business hours. For more detail on 4G failover configuration see 4G failover and backup internet.

QoS — Prioritising VoIP Traffic

Quality of Service (QoS) is a mechanism that allows a router or firewall to prioritise certain types of network traffic over others. With QoS configured correctly, VoIP packets are guaranteed their required bandwidth even when the internet connection is otherwise saturated.

Without QoS, your router treats all traffic equally. If a staff member's computer is performing a large cloud backup at the same time as three calls are in progress, all of that data competes for the same upload bandwidth. The backup is not time-sensitive — a few seconds of delay is irrelevant. The VoIP calls are critically time-sensitive. Without prioritisation, the backup traffic can delay VoIP packets enough to cause audible quality degradation.

QoS resolves this by classifying VoIP traffic (typically identified by its port numbers and protocol) and placing it in a high-priority queue. The router ensures VoIP packets are sent first, before queued data from lower-priority applications. The backup still completes — it just yields bandwidth to the phone calls when required.

QoS is configured at the router or firewall level, not within the phone system itself. This is one of the reasons a business-grade router matters for VoIP deployments. Consumer routers — including many NBN modem-routers provided by residential ISPs — either lack QoS functionality entirely or implement it in a simplified form that is insufficient for mixed-use business environments. A business-grade router from a vendor such as Cisco, Fortinet, Ubiquiti, or similar will have full QoS capability. See business router and firewall for guidance on what to look for.

For businesses on shared internet connections — where multiple tenants share a connection in a building or complex — QoS within your own equipment only controls your local traffic. Congestion upstream, in the shared connection itself, cannot be addressed by your router. This is where the distinction between a shared and dedicated internet connection becomes relevant; see dedicated vs shared internet for more on that topic.

The NBN Upload Speed Problem for VoIP

VoIP is symmetrical by nature. Your voice leaves your office as upload traffic. The other party's voice arrives as download traffic. The bandwidth requirement applies in both directions equally.

This creates a specific problem for NBN services, which are asymmetric by design. Standard NBN plans deliver significantly more download capacity than upload. A 100/20 Mbps NBN plan provides 100 Mbps download but only 20 Mbps upload. For general internet use — browsing, streaming, cloud application access — this is fine, because most internet activity is download-dominant.

For a business running a VoIP phone system alongside video conferencing, the upload figure deserves close attention.

Consider a practical example. A 15-person office has 8 concurrent VoIP calls and 5 staff members on Microsoft Teams video calls simultaneously. The upload requirement for this scenario is approximately:

• 8 VoIP calls × 100 Kbps = 0.8 Mbps for voice
• 5 Teams video calls × approximately 1.5 Mbps per call (outgoing video stream) = 7.5 Mbps
• Cloud application sync, email sending, and background updates: 2–5 Mbps

Total upload requirement: approximately 10–13 Mbps, potentially peaking higher.

A 100/20 NBN plan provides 20 Mbps upload — leaving only 7–10 Mbps of headroom above the baseline requirement. In practice, upload utilisation may peak above the average figures used above, and during those peaks the upload pipe becomes congested. VoIP packets experience jitter and delay. Call quality degrades.

The solution may be a higher-tier NBN plan with greater upload capacity, a move to NBN Enterprise Ethernet which offers symmetric speeds, or an upgrade to a dedicated fixed-line connection. For a comparison of options see NBN business vs enterprise ethernet and upload vs download speed for business.

The upload problem is also why it is worth reviewing the NBN business vs residential internet distinction — business NBN plans often offer higher upload tiers and additional guarantees that make a material difference in mixed VoIP and video environments.

What to Do When VoIP Quality Is Poor

Poor VoIP quality can have multiple causes, and diagnosing the right one before making changes avoids wasted effort. Work through the following steps.

Step 1 — Run a VoIP-specific quality test

A standard speed test is insufficient. It measures throughput but not latency, jitter, or packet loss under realistic conditions. Use a VoIP-specific tool such as PingPlotter, VoIPmonitor, or your phone system provider's built-in diagnostics. These tools will report latency, jitter, and packet loss to the VoIP infrastructure you are connecting to, giving you the three metrics that actually determine call quality.

Step 2 — Test during business hours, not off-peak

This is a critical point that is frequently overlooked. An internet connection that performs well at 7:00pm will often perform differently at 10:00am on a Tuesday. Congestion — both within your local network and in the upstream network — occurs during business hours. If your VoIP tests show acceptable quality in the evening but call quality is poor during the day, the problem is congestion-related, not a fault with your equipment.

This type of congestion may relate to the contention ratio of your internet service — the ratio of subscribed bandwidth to actual available capacity in the shared network. A high contention ratio means more users competing for the same infrastructure during peak periods.

Step 3 — Check upload utilisation during problem periods

Use your router's traffic monitoring, or a network monitoring tool, to observe upload utilisation during the times when call quality is poor. If upload utilisation is running at or near 100% when calls are dropping, bandwidth exhaustion on the upload path is the likely cause. The solution is either a higher upload tier, QoS configuration, or both.

Step 4 — Verify QoS is configured correctly

Confirm that your router has QoS enabled and that VoIP traffic is correctly classified and prioritised. If you have recently changed routers or had your router reset to factory defaults, QoS configuration may have been lost. Check the configuration against your router's documentation or contact your IT support.

Step 5 — Isolate the problem by time of day

If the problem occurs consistently at specific times — typically mid-morning and early afternoon when business internet usage peaks — this points to congestion rather than a hardware fault. Congestion that is localised to your building or network will be addressable through QoS and bandwidth upgrades. Congestion in the upstream network may require a change of provider or connection type.

If the problem occurs at all times of day, equally, the cause is more likely to be a configuration issue (incorrect SIP settings, codec mismatch), a hardware fault (failing router or switch), or a structural limitation of the connection technology (such as a degraded copper segment on an FTTN connection).

How Pickle Supports VoIP and Cloud Phone Systems

Pickle provides business telecommunications and cloud phone systems to Australian SMBs, with internet plans sized and configured for VoIP workloads. This includes selecting connection types with appropriate latency characteristics, configuring upload speeds that account for mixed VoIP and video environments, and ensuring business-grade routers are in place to support QoS.

If you are setting up a new cloud phone system and want to make sure your internet connection is correctly sized, or if you are experiencing call quality issues on an existing system, Pickle can assess your current setup and recommend the right path forward.

Call 1300 688 588 or email [email protected] to speak with the team.

Frequently Asked Questions

Q: How much internet bandwidth does a VoIP phone call use?

A: A single VoIP call typically uses between 26 Kbps and 87 Kbps depending on the codec in use. G.711 and G.722 (the most common codecs in Australian cloud phone systems) use approximately 87 Kbps per call including packet headers. G.729 uses approximately 26 Kbps in constrained environments. For planning purposes, allow 100 Kbps per concurrent call to include overhead and provide a safety margin.

Q: Why does my VoIP call quality get worse during business hours?

A: This is almost always a congestion problem. During business hours, more users on your local network and on the upstream internet infrastructure are active simultaneously. This congestion increases latency and jitter — both of which directly degrade VoIP quality. If your call quality is good in the evening but poor during the day, the solution is typically to configure QoS on your router to prioritise VoIP traffic, to upgrade to a higher upload tier, or to investigate whether your internet provider's network has high contention during peak periods.

Q: Can I use VoIP over a 4G connection?

A: Yes, with caveats. A 4G connection typically delivers one-way latency of 30–80ms, which is within the usable range for VoIP. Calls will often be noticeably less crisp than on a fixed-line connection, and jitter may be higher, particularly in congested mobile cells during business hours. 4G is generally acceptable as a temporary failover path during a fixed-line outage, but it is not recommended as a permanent primary connection for a business relying heavily on VoIP.

Q: What is jitter and why does it affect call quality?

A: Jitter is the variation in the time it takes for consecutive packets to arrive at their destination. Even if your average latency is acceptable, if packets are arriving at irregular intervals — some fast, some delayed — the audio stream cannot be reconstructed smoothly. The result is choppy, robotic, or clipped audio. VoIP devices use a jitter buffer to compensate, but this only works within a limited range of jitter (under approximately 30ms). Beyond that, the buffer cannot compensate without adding unacceptable delay, and audio quality breaks down. Jitter is typically caused by network congestion and is addressed through QoS configuration and ensuring adequate bandwidth during peak periods.

Q: Do I need a special router for VoIP?

A: Not a dedicated VoIP router, but you do need a business-grade router rather than a consumer device. The key requirement is Quality of Service (QoS) support, which allows the router to prioritise VoIP traffic over less time-sensitive data like backups or file downloads. Most consumer routers — including the modem-routers supplied with residential NBN plans — either lack QoS or implement it in a basic form insufficient for business use. A business-grade router from a reputable vendor will provide full QoS capability, along with better firewall features, more reliable performance under load, and proper support for SIP ALG configuration (which should typically be disabled for VoIP to function correctly).