If a VoIP call sounds choppy, delayed, or robotic, the instinct is almost always to check internet speed. That’s usually the wrong place to look. A connection can post a great speed test result and still produce bad calls, because voice quality depends on four specific, measurable network characteristics — bandwidth, latency, jitter, and packet loss — and a speed test only tells you about one of them.
Bandwidth: How Much You Actually Need
Each concurrent VoIP call needs roughly 100 kbps of clear bandwidth for standard-quality audio, or about 150 kbps for HD voice codecs. A 10-person office running calls simultaneously needs that multiplied out, plus roughly 20% headroom on top for protocol overhead, retransmissions, and traffic spikes. In practice, raw bandwidth is rarely the actual bottleneck for a modern business internet connection — the real risk is that voice traffic has to compete with everything else on the network (file syncs, video calls, backups) without any priority, which is a QoS problem, not a bandwidth problem.
Latency: The 150ms Line
Latency is the one-way delay between when a word is spoken and when it’s heard. The ITU G.114 standard puts the acceptable ceiling at 150 milliseconds one-way. Below that, conversation feels normal. Once latency climbs past 200ms, callers start noticing a lag and unconsciously start talking over each other. Past 400ms, holding a normal conversation becomes genuinely difficult. Latency is largely a function of network path and distance to your VoIP provider’s infrastructure, which is why provider selection and network routing both matter, not just your local connection.
Jitter: Why Calls Sound Choppy Even With Good Bandwidth and Latency
Jitter is the variation in delay between packets — not how long they take, but how inconsistently they arrive. Voice systems use a jitter buffer to smooth out small variations, but that buffer has limits. Jitter needs to stay under roughly 15ms for smooth audio; once it consistently exceeds 30ms, the buffer gets overwhelmed and audio quality can degrade quickly and noticeably, even if bandwidth and latency both look fine on paper. Jitter is one of the most common causes of “the call sounds fine sometimes and terrible other times” complaints, because it’s directly tied to momentary network congestion.
Packet Loss: The Least Forgiving Metric
Voice codecs tolerate almost no packet loss. The commonly used G.729 codec needs loss rates far below 1% to avoid audible artifacts, and even a 1% loss rate produces noticeably choppy, dropped-word audio. The generally accepted threshold for acceptable business call quality is under 0.5% packet loss — tighter than most businesses assume, and tight enough that a network issue barely worth noticing for web browsing can make VoIP calls unusable.
Quality of Service (QoS): The Fix Most Businesses Skip
QoS is a network configuration that tells your router and switches to prioritize voice packets over other traffic — so a large file upload or a video call doesn’t crowd out a phone call happening at the same time. This is the single most common gap we find during a VoIP deployment: a business has plenty of raw bandwidth, but no QoS configuration, so call quality degrades unpredictably whenever anything else on the network gets busy. Properly configured, QoS lets voice and data traffic share the same connection without voice quality suffering — dedicating a separate connection to VoIP is one valid approach, but it’s rarely necessary once QoS is set up correctly.
Multi-Location Businesses: Every Site Needs the Same Standard
A phone system is only as good as its worst-performing location. Multi-site businesses commonly get this wrong by provisioning and tuning their headquarters connection carefully while leaving satellite offices on whatever internet plan came with the building. The result is a phone system that works great at HQ and sounds bad everywhere else, on the exact same platform. Bandwidth planning and QoS configuration need to happen at every site individually, not just the main office.
How to Actually Test Your Network
A standard speed test measures throughput — how much data can move per second — and tells you almost nothing about latency, jitter, or packet loss under real conditions. Dedicated network quality testing measures the metrics that actually determine call quality, under conditions that resemble real business use, not a best-case burst test. This is the right first step before troubleshooting a VoIP quality complaint, and before signing with any new VoIP provider.
How CelereTech Approaches VoIP Network Readiness
Before deploying VoIP for a Chicagoland business, CelereTech assesses actual network conditions — latency, jitter, and packet loss, not just a speed test — and configures QoS to give voice traffic priority over competing network activity. After deployment, we monitor call quality on an ongoing basis, so a degrading connection gets caught and fixed before it becomes the kind of persistent problem that makes staff stop trusting their phone system. If you’re evaluating the best VoIP phone system for your business, network readiness is the part most providers won’t ask about until after you’ve signed.
Talk to CelereTech about your network’s VoIP readiness before you deploy or switch providers.