Video Calls Keep Dropping? How to Tell If It Is Your Wi-Fi
Video calls dropping on Wi-Fi? Two ping tests tell you whether the fault is your network, your ISP or the app itself, plus the fixes ranked by effort.
The call freezes for four seconds. Someone says "you cut out," and by the time you have apologized it is fine again. Twenty minutes later it happens twice more.
Nothing in that experience tells you whether the fault is your Wi-Fi, your internet connection, or the app — and the three have completely different fixes. Two ping tests separate them in about five minutes. Here is how to run them and, more importantly, how to read the numbers.
What a call actually needs#
Far less bandwidth than people assume. A one-to-one HD call typically runs on 1.5 to 3 Mbps in each direction. A group HD call wants roughly 3 to 4 Mbps up and 6 to 8 Mbps down, and sharing your screen adds a megabit or two. Set that against a 4K stream at 15 to 25 Mbps and the trap becomes obvious: a speed test reporting 300 Mbps tells you almost nothing about whether your calls will hold together.
What calls are fussy about is timing.
- Round-trip latency. Under about 150 ms is the working ceiling for a call, and under 100 ms feels natural. Past 250 ms, people start talking over each other.
- Jitter — the variation in arrival times between packets. Under 30 ms is comfortable. Past 50 ms the app's buffer runs dry and you get the metallic half-second dropouts.
- Packet loss. Under 1 percent is invisible. Around 3 percent, audio turns robotic. Past 5 percent, calls drop.
Bandwidth is a floor you almost certainly clear. Jitter and loss are what actually break meetings. The full budget for a desk you use every day, including why the upload side runs out first, is in the home office Wi-Fi setup.
Wi-Fi, ISP or the app: the signatures#
Each culprit fails with a recognizable pattern. Read down the column that matches what you notice.
| What you notice | Wi-Fi | ISP or upstream | App or far end |
|---|---|---|---|
| Only you freeze; everyone else sounds fine | Likely | Possible | Unlikely |
| Everyone in the meeting freezes at the same moment | No | No | Yes |
| It happens when you stand up or change rooms | Yes | No | No |
| Bad on Wi-Fi, clean on an Ethernet cable | Yes | No | No |
| Bad on Wi-Fi and on Ethernet | No | Likely | Possible |
| Only bad in one room of the house | Yes | No | No |
| Same trouble every evening, every day | Possible | Likely | No |
| Other apps and downloads are also unstable | Yes | Yes | No |
| One app is fine, the other is not | No | No | Yes |
One row settles more than the rest: plug into the router with a cable and take a call. If the wired call is clean, Wi-Fi owns the problem and the rest of this article applies. If the wired call is just as bad, stop reading about placement and work through when the problem is really your ISP instead — no amount of coverage engineering fixes a bad line.
Ruling out the app before you blame the house#
Zoom, Teams and Meet all expose a live connection panel showing latency, jitter and packet loss per stream, split between what you send and what you receive. Open it during a bad moment. If the loss is on the receive side while your own send statistics stay clean, the trouble is upstream of you or at the other end, and nothing you change at home will help. The thirty-second version of the same test: join the same meeting from a phone on cellular and see whether it also stutters.
Two ping tests that settle it#
The logic is simple. Measure the leg you own separately from the leg you rent.
- Find your router's address. It appears as "default gateway" or "router" in your network settings, commonly 192.168.1.1 or 192.168.0.1.
- Ping the router 100 times from exactly where you take calls. On macOS or Linux,
ping -c 100 192.168.1.1. On Windows,ping -n 100 192.168.1.1. Do not walk around during the test. - Read three things, and ignore the average. You want the maximum, the spread between minimum and maximum, and the packet loss percentage. The average hides exactly the events that ruin calls.
- Ping past the router. Repeat for another 100 packets against a public DNS address such as 1.1.1.1. The difference between this result and the first one is your internet connection's contribution.
- Run the router ping again standing next to the router. That is your control. It tells you what your own hardware is capable of when distance is not a factor.
Then match your numbers against these:
| Result | What it means |
|---|---|
| Router ping: 1-5 ms average, max under 30 ms, 0 percent loss | The Wi-Fi leg is healthy. Look upstream or at the app |
| Router ping: low average but maximums of 200-900 ms | Classic weak or congested Wi-Fi. Those spikes are your jitter |
| Router ping: 1-5 percent loss | Wi-Fi. Loss on your own network should be zero |
| Clean beside the router, ugly at your desk | Coverage. The desk is the problem, not the router |
| Router ping clean, internet ping spiky | Upstream: the line, or your own upload saturating it |
| Both clean, calls still break | The app, the far end, or the device |
A healthy 5 GHz link one room from the router answers in 1 to 5 ms with no loss at all. Once you see 100 ms maximums on that first hop, the radio is retrying failed transmissions — and every retry becomes jitter the call has to absorb. If the spikes only appear while something is uploading, you are looking at bufferbloat rather than coverage, which is the same effect that wrecks online games.
The freeze that happens when you stand up#
If calls break when you get up for coffee, or thirty seconds after you settle into a different chair, you are watching a handoff, not a coverage failure.
Devices are stubborn. A laptop or phone typically clings to the access point it first joined until the signal falls to roughly -70 to -75 dBm, long past the point where a nearer radio would serve it better. When it finally lets go, the switch costs anywhere from under a second on a network with fast roaming enabled to five, ten or twenty seconds if the device fully drops the association and re-runs authentication and DHCP. A call does not survive twenty seconds.
Three things help, in this order:
- Take calls from one spot where the link is strong. Unglamorous, and it works.
- Give your nodes real overlap. Neighboring coverage should meet at around -67 dBm, not at -80. That means placing the second radio closer in than feels necessary.
- Prefer fewer, better-placed radios over more of them. Every extra access point is another candidate your laptop can choose badly.
The mechanics behind all three — sticky clients, the roaming standards, and how to design the overlap — belong to why Wi-Fi drops when you walk between rooms. The case worth flagging here is the desk that sits at the edge of both the router's and the extender's coverage, where a laptop can bounce between them mid-call. If the room is generally slow as well as unstable, start with the one-room diagnostic.
Fixes, ranked by effort#
Work down this table and stop when calls hold. The first four cost nothing.
| Effort | Fix | What it typically buys |
|---|---|---|
| 0 min | Take calls from the side of the room nearer the router, door open | 5-15 dB |
| 2 min | Laptop off the floor, out of the cabinet, away from the microwave and metal shelving | 2-8 dB |
| 5 min | Put the call device on 5 GHz and leave 2.4 GHz to the doorbell | Lower jitter, fewer retries |
| 5 min | Pause cloud backup and large uploads during calls | Fixes upload starvation |
| 15 min | Move the router out of the corner, off the floor, out of the closet | 5-15 dB everywhere at once |
| 30 min | One Ethernet run to the desk | Removes Wi-Fi from the equation |
| 30 min | Extender or mesh node on the near side of the worst wall | 10-20 dB in that room |
| 1-2 hr | Coax or powerline backhaul to a second access point | A backhaul that is not sharing airtime |
Those top rows stack. A laptop on the floor in the far corner of a back bedroom can sit 10 dB or more below the same laptop on a desk near the doorway, and 10 dB is the difference between a link that retries constantly and one that does not. The free rows are not consolation prizes; often enough they are the entire fix.
When an extender makes video calls worse#
An extender is a repeater. It listens to the router and says it again, and for real-time traffic that carries two costs.
Latency. Each hop adds delay — usually a couple of milliseconds on a good dual-band unit, considerably more when the backhaul is marginal and retries stack up. On its own, survivable.
Airtime. A single-radio extender repeats every packet on the same channel, roughly halving usable throughput and doubling the airtime your call consumes. In an already busy band that is exactly the wrong direction, and it can turn an occasional freeze into a constant one.
So an extender is a reasonable answer when your office reads somewhere around -72 to -80 dBm and there is an outlet in between where the router's signal is still strong. It is the wrong answer when the real problem is a congested channel, when the office sits behind concrete with no healthy midpoint, or when the desk never moves — because in that last case a single cable is cheaper, faster and permanently boring. The rule that decides the good case is in choosing the outlet an extender plugs into: halfway in signal, not halfway in feet.
Plan the desk before you buy the box#
Only one question in this article has to be answered before you spend money: how much signal does the chair you take calls from actually get, and would a second radio change that? Range Up works it out on a plan rather than by moving hardware around. Draw the rooms, set the material preset to what the house is really built from, mark where the router stands today, and read the predicted level at the desk. Name that room, and the app sorts the extender and node positions it can try by how much each raises the level at that desk, then draws the best of them over your coverage as it stands today.
It is a simulation that runs on the phone, so it never touches your live network — jitter and packet loss stay yours to measure with the pings above. Coverage is the half it does settle, and the negative result counts for as much: if the best candidate position moves the desk by only three or four dB, the office does not need another radio, it needs one of the wired rows in that table. Put the office on a floor plan before the box arrives. The guide to positioning a second radio covers where it has to go once you know you need one.
Frequently asked questions#
Why do my video calls drop when my speed test says everything is fine?
Because a speed test measures peak throughput over several seconds, while a call depends on packets arriving evenly every few milliseconds. A connection can average 300 Mbps and still deliver half-second gaps caused by retries, congestion or a roaming handoff. Test latency, jitter and packet loss with a sustained ping instead of a speed test.
How can I tell if the problem is Zoom or Teams rather than my network?
If everyone in the meeting freezes at the same moment, the fault is the service or the other participant, not your house. Open the app's connection statistics during a bad stretch: loss on the receive side with clean send statistics points away from your network. Joining the same meeting from a phone on cellular confirms it in under a minute.
What jitter and packet loss numbers are too high for a video call?
As rules of thumb, jitter under 30 ms and packet loss under 1 percent are comfortable, jitter past 50 ms or loss around 3 percent produces choppy audio, and loss above 5 percent drops calls. Round-trip latency wants to stay under about 150 ms, feels natural under 100 ms, and past 250 ms makes conversation awkward. On your own network, between a device and its router, packet loss should be zero.
Will a Wi-Fi extender stop my calls from dropping?
Only if the cause is weak coverage at your desk and there is a spot between the router and that room where the router's signal is still strong. An extender adds a hop and, on single-radio models, consumes twice the airtime for the same traffic, so it can make calls worse when the real problem is a congested channel or interference. Diagnose the cause before buying one.
Why does my call freeze when I walk into another room?
Your device is switching access points, or refusing to switch until the old signal is nearly unusable. That handoff costs anything from a fraction of a second to twenty seconds, depending on whether the network supports fast roaming and whether your device drops the connection completely. Taking calls from one spot, and placing nodes so their coverage meets at around -67 dBm rather than -80, both help.
Is an Ethernet cable really worth it for a home office?
For a desk that does not move, yes — it removes coverage, interference, airtime and roaming from the list of suspects in one step, which is why it sits at the bottom of the table above as the fix that ends the argument. A single run along a baseboard takes about half an hour. If running a cable is impossible, coax or powerline backhaul to a second access point gets you most of the benefit.
Does turning my camera off actually help?
Sometimes, because it cuts your upload requirement by a few Mbps, which matters if your upstream is genuinely saturated or your plan has a thin upload. It does nothing at all for jitter, packet loss or roaming problems, which are the more common causes of a dropping call. If turning the camera off fixes things reliably, look at your upload capacity rather than your Wi-Fi coverage.