VoIP Call Quality Problems: 8 Network Issues Your IT Provider Should Have Fixed Already

The calls are embarrassing and you are out of ideas. A client hears every third word. A voice turns robotic mid-sentence, recovers, then drops entirely. You have rebooted the phones, swapped a handset, and sat through the ritual where the phone vendor blames the internet, the internet provider runs a speed test and declares everything fine, and nobody fixes anything.

Here is what that loop keeps missing: VoIP call quality problems are almost never phone problems. A handset has one job — turn your voice into a stream of data packets and play back the packets it receives. When audio is choppy, robotic, one-way, or gone, the phone is faithfully playing a damaged stream, and the damage happens in between: switches, Wi-Fi, router, firewall, internet circuit. NSN Management installs and supports business phone systems across Tulsa as a 3CX Gold Partner, and nearly every quality complaint we have traced ended at one of the eight network issues below — all measurable, all fixable, and all things a competent IT provider handles before you notice them.

Why call quality lives in the network

A VoIP call is a real-time stream: a small packet of audio leaves roughly every 20 milliseconds — about fifty per second in each direction — and each one has to arrive on time and in order to be useful. Email, web pages, and file syncs have no such deadline — a lost packet is quietly re-sent and nobody notices. Voice has no time for that: a packet that arrives late might as well not arrive at all.

That deadline is why a network that feels fine for everything else can still butcher a phone call — and it is also the good news: jitter, latency, and packet loss are numbers, not mysteries, and each of the eight problems below leaves fingerprints in them.

1. No QoS: your voice waits in line behind everything else

The symptom: Calls are fine at 7:30 a.m. and rough by mid-morning. Quality collapses whenever a video meeting, cloud backup, or big upload starts.

The cause: Out of the box, most networks treat every packet identically — first come, first served. Voice queues behind file syncs, video streams, and software updates, and when the line fills, voice packets arrive late or not at all. The applications filling the line do not care about a half-second delay; a phone call does.

The fix: Quality of Service — tagging voice packets (typically DSCP EF) so switches and the firewall move them to the front of every queue, putting phones on a dedicated voice VLAN, and shaping traffic so the circuit never quite saturates. None of it is exotic — it is standard configuration for day one of any phone deployment.

2. Jitter: packets arriving out of rhythm

The symptom: Voices turn robotic, underwater, or stuttery — words smear together or fracture into syllables.

The cause: Packets leave every 20 milliseconds, but congestion makes them arrive irregularly — three bunched together, then a gap. That variation is jitter. Every phone has a jitter buffer that holds packets briefly to restore the rhythm, and it hides small variation invisibly. It cannot hide large or spiky jitter: late packets get discarded, which you hear as garble, or the buffer grows deep enough to add noticeable delay. A jitter buffer is a shock absorber, not a fix for a bad road.

The fix: Measure jitter at the network edge — a common rule of thumb is to keep it under about 30 milliseconds — then remove the cause, which is nearly always congestion that QoS should be handling.

3. Latency: delay beyond about 150 milliseconds

The symptom: You keep talking over each other. Responses land a beat late, and calls take on the rhythm of a satellite interview.

The cause: The long-standing ITU-derived rule of thumb is that one-way delay beyond roughly 150 milliseconds becomes noticeable in conversation. Delay accumulates quietly: a saturated uplink holding packets in deep queues, voice backhauled through a VPN to a distant office, a firewall running heavy inspection on real-time traffic, or a phone platform hosted far from Oklahoma.

The fix: Measure round-trip time to the phone platform, route voice directly to the internet instead of through VPN detours, keep queues shallow with traffic shaping, and choose platform regions close to home.

4. Packet loss: even 1% is audible

The symptom: Syllables clip. Words vanish. Callers say "you cut out" and you find yourself repeating sentences.

The cause: Lost packets come from saturated links, damaged patch cables, failing switch ports, duplex mismatches, and Wi-Fi interference. Modern codecs conceal a single lost packet well, but losses cluster in bursts, and bursts are audible. The rule of thumb: around 1% packet loss is audible, and a few percent makes conversation genuinely difficult — even while web browsing on the same connection feels normal.

The fix: Localize the loss — inside the building or out on the circuit — with per-segment testing, then fix the physical cause. And monitor continuously: packet loss should be an alert your provider sees, not a complaint your receptionist collects.

5. SIP ALG: the router 'helper' that breaks calls

The symptom: One-way audio — you hear them, they cannot hear you, or the reverse. Calls that drop at the same minute mark. Phones that randomly unregister. Inbound calls that go straight to voicemail while the desk phone sits silent.

The cause: SIP ALG (Application Layer Gateway) is a router feature that rewrites call-signaling packets, intended to help VoIP traffic pass through NAT. In practice, implementations are broken often enough that phone platform vendors almost universally say to disable it. It mangles packets, breaks registrations, and produces exactly the intermittent failures above — and it ships enabled by default on a great many routers, including the ones internet providers hand out.

The fix: Turn SIP ALG off, then configure the firewall the way the phone platform actually specifies — consistent NAT behavior and registration keepalives. This is one of the first things a competent installer checks. If nobody has ever looked at yours, this paragraph may have just found your problem.

6. Double NAT and consumer-grade routers

The symptom: Intermittent one-way or missing audio. Phones that show "registered" but do not ring. Problems that vanish when you test a call over a cell phone's hotspot.

The cause: Double NAT is what you get when the internet provider's gateway acts as a router and your own router sits behind it doing the same job. SIP signaling carries network addresses inside the packets themselves, and two layers of address translation reliably confuse it. Consumer-grade routers compound the trouble: SIP ALG on by default, small connection tables, no voice VLAN, no real QoS.

The fix: Put the provider's gateway into bridge mode so exactly one device routes, and make that device a business-grade firewall sized for your headcount — with VLANs, QoS, and a documented configuration someone can actually support.

7. Phones on congested Wi-Fi instead of wired connections

The symptom: Quality depends on which room you are in. Laptop softphones sound worse than desk phones. The conference room is fine at 8 a.m. and useless once the building fills up.

The cause: Wi-Fi is shared airtime: every device in range takes turns transmitting, and interference forces re-sends that show up as jitter and loss. A phone competing with laptops, video meetings, and personal devices loses that contest often enough to hear it. Roaming between access points mid-call adds its own stumble.

The fix: Wire what does not move. Desk phones belong on Ethernet — powered over the cable, on the voice VLAN, prioritized end to end. Where wireless is genuinely needed, business-grade access points with voice prioritization and a deliberate channel plan make Wi-Fi calling workable. Putting desk phones on the same Wi-Fi as everything else is a choice a competent provider does not make.

8. An undersized circuit — and no failover when it fails

The symptom: Everything degrades during busy hours. And when the internet blips, the phones do not degrade — they die, and callers hear dead air or endless ringing.

The cause: A single call needs only about 100 kilobits per second in each direction, as a rule of thumb, so headline bandwidth is rarely the real issue. The traps are subtler: asymmetric circuits whose thin upload side saturates first — your callers hear the damage before you do — and a cloud phone system riding a single circuit, which turns every internet outage into a phone outage.

The fix: Size the circuit for real concurrent usage with attention on upload, then add a failover circuit from a different carrier on a different path — and test that calls actually survive the switchover, because failover that has never been tested is a hope, not a design.

Match the symptom to the likely cause

How to tell which one you have

You can narrow the diagnosis before anyone opens a router, just by noticing the pattern:

• When does it happen? Bad at busy hours and fine early or late points to congestion and missing QoS. Bad all the time, regardless of load, points to configuration — SIP ALG, double NAT.
• Who does it happen to? One desk suggests a cable or switch port. Everyone at once suggests router, firewall, or circuit. Only wireless devices suggests Wi-Fi.
• Which direction fails? One-way audio is the signature of SIP ALG and NAT trouble, not of a slow connection.
• Is it eerily regular? Calls dropping at the same minute mark are timeouts, not bad luck.
• The hotspot test. If a call over a cell phone's hotspot sounds fine from the same desk, the problem is inside your network — good news, because that is the part someone can fix.

Patterns narrow it; measurements confirm it. Modern phone platforms log jitter, loss, and latency for every call — 3CX keeps per-call quality statistics, for example — and edge testing pins down which segment is at fault. You should not have to run any of this yourself, but knowing it exists changes the conversation: "the phones are bad" invites a shrug, while "we get one-way audio on inbound calls every afternoon" invites a diagnosis.

What a provider who owns the problem looks like

The blame loop at the top of this article exists because responsibility is split three ways: the phone vendor owns the handsets, the internet provider owns the circuit, and somebody — often nobody — owns the network in between. A call crosses all three, so each party can truthfully say their piece works. The fix is structural: one provider who owns the path end to end. Here is what that looks like:

• Assessment before installation. Circuit quality tests, a firewall and router review, cabling checks, and a Wi-Fi survey — before a phone system is quoted, not after it disappoints.
• The unglamorous checklist, done up front. SIP ALG off, voice VLAN built, DSCP tagging and priority queues configured, firewall rules matched to the platform's specification, failover tested with real calls.
• Monitoring after go-live. Per-call quality statistics reviewed, alerts on jitter and loss, problems investigated before people mention them.
• One number to call. When something does break, the provider works the carrier ticket for you instead of handing you a phone tree. Response standards are agreed up front and reported monthly.

This is how NSN Management approaches IT support in Tulsa: phones are not a gadget bolted onto the network; they are its most demanding application, and they get managed that way. Tulsa-owned since 2012 and a 3CX Gold Partner, we build and support the whole path, from handset to carrier. Phone systems usually fold into a broader support agreement rather than standing alone; our guide to managed IT services cost in Tulsa lays out what that typically looks like.

None of the eight issues above is rare, and none is unfixable — each has a known cause, a measurable signature, and a standard remedy. If your team has been apologizing for the phones, stop apologizing and get the network diagnosed: book a Discovery Call with NSN Management, or call 918-770-9150, and we will trace your call quality problems to their cause, show you the measurements, and tell you plainly what fixing them takes.