Mobile data or Wi-Fi: which one actually holds a video call together

The instinct is that Wi-Fi is the serious connection and mobile data is the emergency one. For downloads that is usually true. For a live conversation it often is not, because a call does not want a big pipe — it wants a small, boring, uninterrupted trickle that arrives on time. A connection that can pull a film in four minutes and then stalls for half a second every so often is worse, for this purpose, than a modest one that never stalls at all.

Which means the honest answer depends on your building, your router, your neighbours and the hour of the evening — and it is settled by testing rather than arguing. Here are the questions people ask when their picture keeps freezing, in the order they usually ask them.

Which one is actually steadier for a video call?

Whichever one is having a quieter night, which is genuinely not a dodge. Good Wi-Fi close to the router beats almost any mobile signal, because the trip is short and nothing is competing for it. But home Wi-Fi degrades quickly with distance, walls and the number of devices hanging off it, and in a flat surrounded by other flats it is sharing the airwaves with dozens of networks you cannot see. A solid mobile signal is often more consistent than third-bedroom Wi-Fi in a converted terrace. The pattern that holds nearly everywhere: same room as the router, use Wi-Fi. Far from it, on a busy network, or anywhere public, try mobile data.

Why does a fast Wi-Fi connection still stutter?

Because speed and steadiness are different measurements and a call only cares about the second one. A speed test measures how much you can move in a burst when the connection is given the whole road. A call needs a much smaller amount of data delivered every few milliseconds without fail, and that is ruined by jitter — packets arriving late or out of order — and by short queues that build up when something else on the network suddenly wants bandwidth. That is why your call can freeze at the exact moment a phone in another room starts a backup, on a line that benchmarks beautifully.

How much does distance from the router matter?

More than almost anything else you can change for free. Wi-Fi loses strength fast with distance, and every solid thing in the way — brick, concrete, foil-backed insulation, a mirror, a fridge, a bathroom — takes another bite. A router two rooms away behind a chimney breast can be a fraction of the connection it is in its own room, even though your phone still shows bars. Bars measure signal, not usable throughput, so they are optimistic. Before you buy anything or change provider, try one call sitting near the router. If that call is clean, your problem is not your broadband; it is the distance between you and the box.

Is 5 GHz better than 2.4 GHz for calls?

Close to the router, yes, clearly. The 5 GHz band is faster, and far less crowded because fewer devices and fewer neighbours use it. Its weakness is range: it fades faster and walls hurt it more, so at the far end of a flat the older 2.4 GHz band can be the steadier choice even though it is slower on paper. Most routers broadcast both, sometimes as two network names and sometimes merged into one that decides for you. If you get two names, it is worth trying each from the seat you actually use, because the winner depends on your walls rather than on which number is bigger.

How much data does an hour of video chat use?

There is no single honest number, and anyone who gives you one is averaging over hardware, resolution, lighting and how much movement is in the picture — a still person in a dim room costs far less to encode than someone waving in daylight. Measure your own instead. Note the figure in your phone's data usage screen, take one call, then look again: the difference is your real cost on your device and your network. Do it once and you will know whether this fits your plan, which is more useful than any figure printed on a page written by someone who has never seen your phone. What is safe to say without measuring anything is that the camera dominates the total: a voice call costs a small fraction of a video one.

Does mobile data drain the battery faster than Wi-Fi?

Generally yes, and it gets worse the weaker the signal, because a phone with a poor connection turns its radio up to shout. That is why an hour on one bar in a basement costs far more battery than an hour on full signal outdoors. Wi-Fi is usually gentler, though not if you are at the edge of range where the same shouting problem applies. The camera and the screen typically cost more than either radio, so if battery is the real constraint, taking the voice lane does more for you than switching networks. A hot phone also throttles, which quietly makes the picture worse.

Should I turn the camera off to save data?

Dropping video is the single most effective thing you can do, because video is most of the cost of a call — but do it by starting a voice match, not by covering the camera mid-call. The cover is a privacy screen the other person sees; your camera keeps running behind it, so the data does not stop. A voice match never opens the camera at all. On a metered plan or a marginal signal that is not a compromise, it is the sensible default — most people find the conversation barely changes, and the ones where it does are the ones where you would have wanted the camera anyway. Voice only versus camera on goes through what each lane costs you socially rather than technically.

Is one bar of mobile signal better than weak Wi-Fi?

Sometimes, and the only way to know is to try both. Bars are a crude measure of signal strength and tell you nothing about how many other people are on the same mast, which is often the real limit in a busy area at a busy hour. A single bar that nobody else is using can carry a call perfectly; three bars in a packed stadium cannot carry anything. The same is true of Wi-Fi in a block of flats at nine in the evening. Test rather than reason about it: one match on each, ten minutes apart, and believe whichever one let you finish a sentence.

Why do calls fall apart on café and hotel Wi-Fi?

Two separate reasons, and they need different responses. The first is sharing: everybody in the building is on one connection, and evenings are when everybody is streaming. The second is the network itself — many public networks put you behind a sign-in page, throttle you after a while, or restrict the kind of traffic a live call needs, which is why a call can connect and then die a minute later while web pages keep loading fine. If a public network fails once, do not spend twenty minutes fighting it. Switch to mobile data, or take a voice match, which asks far less of a bad connection.

Does a VPN make the picture worse?

Usually a little, occasionally a lot, and very rarely it helps. A VPN sends your traffic on a detour through another server, and a detour adds distance, which adds delay — how much depends entirely on where that server is. A nearby one may cost you almost nothing; one on another continent adds a lag you will feel in every pause. The rare case where it helps is when your own network is doing something clumsy to live traffic and the tunnel hides it. If calls are poor and a VPN is on, turn it off for one match and compare. That is a thirty-second experiment.

Is it worth switching networks in the middle of a call?

Rarely. The switch itself drops the connection for a few seconds, and in a random-match conversation those seconds usually end the call anyway. It is better to decide before you start, or to fix it the cheap way mid-call by turning the camera off and staying on audio. Switch between matches instead, when it costs nothing: turn Wi-Fi off, start the next conversation on mobile data, and see whether the evening improves. The one exception is if you are walking out of range of your own Wi-Fi, where switching early beats being dropped by the doorway — phones cling to a dying Wi-Fi signal for far longer than they should before giving up and reaching for mobile data.

Why does it get worse at the same time every evening?

Because everybody around you is home. Shared connections have peaks, and mid-evening is the obvious one for home broadband and for mobile masts in residential areas alike. If your calls are fine at seven in the morning and unusable at nine at night, congestion is your answer, not your equipment, and no amount of restarting the router changes that, however satisfying it feels to try. Your practical options are to use the band or network that is less crowded, to drop to voice during the peak, or to talk at a different hour. It is one of the few connection problems where nothing is broken.

Can the other person's connection be the problem?

Frequently, and it is the possibility people forget. Every call has two ends and either can be the weak one. The tell is asymmetry: if they freeze while your own picture stays smooth and everything else on your phone is quick, their side is struggling. There is nothing you can do about it and nothing to feel bad about — on random matching the fix is to say so plainly, and if it does not clear up, skip. Some of the freezes people blame on their own network are somebody else's train going into a tunnel on the other side of the world. Before you spend an evening diagnosing your own router, notice whether the trouble follows you from one conversation to the next or arrives with particular people.

What can I change in ten seconds when the picture breaks up?

In rough order of how much they help: cover your camera and continue on voice; move nearer the router or towards a window; stop anything downloading on your own device; and if you are on public Wi-Fi, drop to mobile data. Those four fix most of it. What does not help in ten seconds is restarting the router, running a speed test, or explaining to the other person what your internet provider promised you. If the call is still unusable after those, end it and start the next one — random matching means a fresh connection is a tap away, which is a luxury a scheduled call does not have.

Do downloads, TVs and other people in the house really matter?

Yes, and more than most people believe. Home connections are shared, and the pattern that hurts a call is not steady streaming but sudden bursts — a console updating, a phone backing up photos the moment it joins the Wi-Fi, a laptop deciding to install something. Those bursts fill the queue for a second or two and a second or two is a frozen face. If your calls break up unpredictably in a busy house, that is the likely cause. Some routers can prioritise one device, but the simpler fix is to take the call on mobile data and leave the household to it.

Does a hot phone mean a bad connection?

It usually means the opposite of what people assume: the phone is working hard, and heat is the consequence rather than the cause. Once it gets hot enough it throttles to protect itself, and a throttled phone encodes video worse, which looks exactly like a network problem. Weak signal makes this worse because the radio works harder. Take the case off, unplug the charger if you are also charging, and drop to voice for a few minutes — the phone cools and the picture usually recovers on its own. If your calls get steadily worse over twenty minutes rather than randomly, heat is the first suspect.

What about trains, cars and walking around?

Moving between masts costs you a break every time, and trains are the worst case: cuttings, tunnels and hundreds of people on the same connection. Video is close to hopeless on a typical journey and voice is often fine, which is the practical answer rather than a consolation prize. If you are a passenger and want a conversation, take the voice lane, use headphones, and expect to lose it a couple of times — a dropped conversation on a train is the network, not you, and the next match is a tap away. Do not do any of this while driving. And remember what your camera shows if you are outdoors — a street sign or a station board tells a stranger exactly where you are.

What is the honest test for my own home?

Take two conversations back to back in the same seat within the same hour: one on Wi-Fi, one with Wi-Fi switched off so the phone uses mobile data. Do not measure anything. Just notice how often the picture froze, whether the sound broke up, and whether you finished sentences. Repeat it once at your usual hour and once at a quiet hour, and note down which network you were on each time, because after three conversations nobody remembers. Congestion changes the answer. Ten minutes of this beats any amount of reading, because it tests your building, your router, your network and your phone together — which is the only combination that matters.