Business desk phone connected to a cloud phone system

G U I D E

Dropped Calls on Cloud Phones: Causes and Fixes

How to diagnose dropped calls, one-way audio and poor voice quality on a cloud phone system.

By Jarrod Lilford, Director/Owner, Kookaburra Comms · Last updated:

Dropped calls and choppy audio on cloud phone systems are almost always caused by the local network or internet service rather than the phone platform. The usual causes are insufficient or unstable upload bandwidth, jitter and packet loss, no traffic prioritisation, Wi-Fi handover problems, or a firewall closing sessions early. Diagnose the path before changing providers.

Key facts

The short version

When calls drop on a cloud phone system, the instinct is to blame the provider. In practice the fault is usually somewhere in the path between the handset and the internet, and it is almost always about consistency rather than capacity.

Voice uses very little bandwidth. What it cannot tolerate is variation. Packets arriving late, out of order, or not at all produce the symptoms people report as choppy audio, robotic voices, one-way audio and dropped calls. Each symptom points at a different part of the path.

Symptom Most likely cause
Choppy or robotic audio Jitter or packet loss on the internet service
One-way audio Firewall, NAT or routing problem
Calls drop at a consistent time Firewall session timeout
Audio degrades at busy times No traffic prioritisation, saturated upload
Problems only on cordless or mobile Wi-Fi roaming and coverage
One handset only Cabling, switch port or the device itself

Establish where the problem is

Start by narrowing the scope. Does it affect every user or one? Every call or only external calls? All day or only at peak times? Handsets, softphones, or both?

A problem affecting one handset is a device, cable or switch port issue. A problem affecting everyone at 11am and 3pm is contention. A problem affecting only softphones on laptops points at Wi-Fi. A problem affecting only external calls points at the internet path or the firewall.

Record the time and the number of a few affected calls. Most cloud platforms retain call quality statistics per call, including jitter, latency and packet loss. That data turns a vague complaint into a measurable one.

Upload is usually the constraint

Business internet services are often sold on download speed, but a phone call is symmetrical. Every person speaking is sending audio out. On a service where upload is a fraction of download, the upstream path saturates long before anyone notices a download problem.

This is why calls degrade when someone starts a cloud backup, uploads a large file, or a video meeting starts in the boardroom. Download-heavy activity may be fine while the upload path is full.

Check the service’s actual upload capacity against the number of concurrent calls plus video meetings, then consider whether the connection type still suits the business. NBN business fibre vs Enterprise Ethernet vs 5G covers where each option fits.

Prioritise voice on the network

If voice shares a connection with everything else and nothing tells the network which traffic matters, voice will lose. Quality of service marks voice traffic so the gateway sends it first when the link is busy.

This needs to be configured in two places to be useful: on the local gateway or firewall, and in a way the internet service respects on the upstream path. Local prioritisation alone still helps considerably, because the most common congestion point is the upload queue on your own connection.

Separating voice onto its own VLAN makes this practical. It also stops a guest device or a backup job from competing directly with a call, and makes the traffic easy to identify when troubleshooting.

Check the firewall before blaming the provider

Calls that drop after a predictable interval — often around 30 or 60 seconds, or consistently at the same duration — are a strong signal of a firewall or router closing the session. Voice sessions need to stay open for the length of the call, and some devices treat an idle-looking session as finished.

One-way audio is the other classic firewall symptom. The call connects, signalling works, but the media path only completes in one direction. This is usually NAT handling or a rule that permits signalling and not media.

Consumer-grade routers are frequent offenders here, particularly those with “SIP helper” or “SIP ALG” features that rewrite traffic in ways cloud platforms do not expect. On a business network the gateway should be a business device with predictable behaviour.

Wi-Fi is a common cause for softphones and cordless

Desk handsets should be cabled wherever possible. Voice on Wi-Fi introduces roaming decisions, interference and airtime contention to a service that depends on consistency.

Where wireless voice is unavoidable — softphones on laptops, mobile apps, cordless handsets — the wireless design has to support it. That means overlapping coverage without excessive co-channel interference, sensible transmit power so devices roam rather than clinging to a distant access point, and a network that separates voice from guest and bulk traffic.

If wireless is the suspected cause, work through why is our office Wi-Fi slow first, because the underlying issues are the same.

Have a plan for when the connection fails

Once phones are cloud-based, the internet connection becomes the phone line. A single service with no backup means an outage takes voice with it.

Practical options include a second connection with automatic failover, a gateway that switches paths without manual intervention, and platform-side call forwarding to mobiles if the site drops entirely. Most cloud platforms can redirect inbound calls when a site is unreachable, but that has to be configured before it is needed, not during an outage. Business internet failover and redundancy covers the design.

When it is the platform

Occasionally the platform genuinely is at fault, and that is what published service levels and status pages are for. Check the provider’s status history against your incident times before spending on network changes.

If quality problems persist across multiple sites, multiple connection types and cabled handsets, and the call statistics show clean jitter and loss figures on your side, escalate with the call data rather than a description. Specific call identifiers and timestamps get a far better response than a report that calls are bad.

Helpful starting points

If the whole site suffers at busy times, start with the business internet service and traffic prioritisation. If only wireless devices are affected, start with Wi-Fi and network solutions. If the phone system itself no longer suits how the business works, business phone systems and UCaaS covers what a migration involves.

Frequently asked questions

Why do our cloud phone calls keep dropping?
The most common causes are packet loss or jitter on the internet service, no prioritisation of voice traffic on a busy connection, a firewall or router closing the call session early, or handset roaming problems on Wi-Fi. The phone platform itself is rarely the fault, so diagnose the network path before changing providers.
How much bandwidth does a cloud phone call need?
A single call uses a small amount of bandwidth, typically well under 100 kbps in each direction. The issue is rarely capacity and almost always consistency. Voice is far more sensitive to jitter, latency and packet loss than to raw speed, and upload is usually the constraint.
Should our phones be on Wi-Fi or cabled?
Cable desk handsets wherever possible. Wi-Fi adds roaming, interference and airtime contention to a service that needs consistency. Softphones on laptops and mobiles will use Wi-Fi by necessity, which is why the wireless design and traffic separation matter.

Related services

Related guides

Sources

Need help applying this to your business?

Talk to Kookaburra Comms about how to put this into practice in your environment. Call 03 9008 4199 or send a message.

GET A FREE QUOTE