A dead zone is not where your phone stops hearing the router. It is where the router stops hearing your phone.
In most homes that failure starts on the client’s uplink — not because it always must, but because the phone in your hand is usually the weaker transmitter in the pair. The router keeps shouting. The phone keeps whispering. And somewhere in your hallway, the whisper stops arriving.
This is the third stop on the operator’s diagnostic checklist — Wi-Fi coverage. If you haven’t cleared the wired baseline and the gateway itself, start there instead, because from the couch a coverage hole and a sick line look identical.
What follows is how a field tech actually finds a dead zone, and the order in which the fixes are worth trying.
What a dead zone actually is
A dead zone is rarely a place with no signal. It is a place where the link falls apart while the bars are still showing.
Two radios that can still see each other, but can no longer hold a useful conversation. As conditions degrade, both ends drop down the rate ladder to slower, more robust modulation, and they start retransmitting the frames that didn’t survive. Those retries eat airtime — the same airtime everything else in the house is waiting for. That is how one distant client quietly becomes a device-capacity problem for the whole house.
The connection is technically alive and functionally dead. That is the shape of it: not silence, but a link spending most of its effort repeating itself. Which is also why the number on the box was never going to describe this room. Link rate is not throughput, and in a dead zone the gap between them is at its widest.
The direction everyone measures wrong
Look at the two ends of the link honestly.
A router sits still and stays plugged in. It has room for antennas, for receive diversity, for thermal headroom, and it never has to think about battery life. A phone is designed around a battery and a thin chassis: less power to transmit with, smaller antennas, and a receiver built under constraints the router never faces. A laptop with a good radio, or a router running conservatively, can narrow that gap — this is a tendency, not a law. But as a tendency, it holds in most homes.
So the failure is lopsided. You can still hear the router after it has stopped hearing you.
And the uplink is not the optional half. The acknowledgments your device sends back, your side of every video call, every file you upload, every message you send — all of it depends on the router hearing you. When that direction collapses, the connection feels broken even though your phone is still showing bars.
This is where the standard advice goes wrong. Most phone-based heatmap apps map what the phone receives. Unless the tool is integrated with access-point-side telemetry, it cannot directly show how well the router hears the client — which is the half that breaks first. You get a beautiful color map of only one direction.
Signal bars measure one direction. The link needs both.
The three clues a field tech reads
Forget the icon. There are three things worth looking at, and none of them is a bar count.
Received quality. RSSI, and where your router exposes it, SNR. Field rules of thumb, not standards: around −65 dBm is a stable target for demanding traffic; around −70 dBm is marginal, and performance becomes dependent on the device and the room; below roughly −75 dBm it is often unreliable for video calls, uploads, and anything else with a deadline. A door sensor may sit happily below that line for years. A video call on your work laptop will not.
Selected rate. The MCS or Tx/Rx link rate the radios are running at that moment. Each direction adapts on its own, so read the number as evidence, not as a verdict. But a rate that stays collapsed — while RSSI is weak and retries are climbing — is a strong sign the link has run out of RF margin.
Link effort. Retries and retransmissions. This is the clue that separates a weak-but-clean link from a link that is fighting for every frame, and it is the one that most often explains why a “fine” signal delivers terrible video.
Your router may expose only one or two of these. Some consumer firmware shows per-client RSSI and Tx/Rx link rates; other systems abstract everything into a single “good / fair / poor” badge and hide the rest. Read whatever you have, and read it in the room — not from the couch.
Why the hole is where it is
Dead zones are not random. They are architectural.
The usual suspects are the ones you would guess: poured concrete, brick, tile, metal studs, mirrors, and large appliances. The ones that surprise people are the modern efficiency features they paid extra for. Foil-backed insulation is a conductive sheet running through the wall cavity. Low-E window glass carries a thin metallic coating — the same coating that rejects heat also reflects radio energy. Both behave less like an ordinary wall and more like an RF barrier. An energy-efficient house can be remarkably good at keeping Wi-Fi out of a bedroom, and even better at keeping it off the patio.
Band choice matters here too, but not for the reason it is usually given. The line you will read everywhere — higher frequency, shorter range — overstates the effect. In open air, 6 GHz and 5 GHz are far closer than the marketing implies. What actually separates them is the walls, because higher frequencies lose more energy passing through material, and the power rules each band operates under. The practical consequence is the same either way: a room that is dead on 6 GHz can be perfectly usable on 5 GHz.
The channel-width trap
Wider channels raise the possible PHY rate. They also collect more noise: at a fixed received signal level, doubling the width adds roughly 3 dB of it, which leaves less SNR margin at the edge of coverage.
Whether that margin loss turns into a real coverage penalty depends on how transmit power scales in that band and on that device — and that is a longer story than this post should tell. It is why channel width has its own guide. For dead-zone work, one instinct is enough: narrower can be more reliable at the edge, but not for the same reason on every band.
The ten-minute walk test
You do not need an app. You need a comparison — against yourself.
- Set a baseline. Stand next to the router with the device that actually has the problem. Run a speed test. Write down both numbers: download and upload.
- Move. Go to the bad spot. Same device, same test server, same app. Run it again.
- Read the gap, not the number. Your plan may be asymmetric — a 35 Mbps upload is not a dead zone, it is a cable plan. What you are hunting for is upload falling away from your own baseline, usually well before download does.
- Run it twice. Same spot, a few minutes apart. Large swings often point to changing interference, contention, a roaming decision, or the test path itself. A drop that repeats, every time, against your near-router baseline makes coverage the likelier story. These are clues, not verdicts — which is the whole point of this post.
- Then look at the router. While you are standing in the bad spot, open the client list: which band did the device choose, and what link rate or RSSI is it holding right now?
If your router doesn’t let you separate the bands, don’t rebuild your network just to run a test. Simply record which band the device picked on its own — that choice is data too.
Fixing it, in the order that works
Cheapest and most reversible first. This is the order a field tech would work through, and skipping to the end is how people end up with a drawer full of extenders.
1. Clear the path. Off the floor. Out of the cabinet. Away from the television, the metal shelf, the mirror. Where the box permanently lives is its own subject, but a router sitting inside a media console behind a screen is a router with an extra wall in front of it — and you can fix that in thirty seconds.
2. Tune the radio. On the band that matters for that room, try a narrower channel, and a less crowded one. A narrower channel can recover some reliability at the edge even though the headline PHY rate drops. It is one of the very few settings where turning a number down makes the result better.
3. Add a node — and mind the backhaul. A repeater with no dedicated backhaul radio has to receive and retransmit on the same radio, and devices behind it can see roughly half the throughput as a result. A dedicated backhaul radio avoids that specific same-radio tax. It does not make the hop free — only a cable does that. If you are going this way, mesh versus one strong router is a decision worth making deliberately rather than by default.
4. Wire it. An Ethernet run, or MoCA over the coaxial cable already in your walls. Any wired backhaul turns a compromised node into a real one.
The one thing everyone gets wrong
An extender does not belong in the dead zone. It belongs in the last place that still has a solid link.
Put it in the bad room and it will do exactly what you asked: faithfully rebroadcast a bad connection. It cannot clean up what it did not receive cleanly. The node’s job is to stand at the edge of good coverage and push that edge outward — which means its link back to the router has to be strong, even when that placement feels like it isn’t far enough into the problem.
If the node cannot get a good link home, nothing behind it will either.
When Wi-Fi isn’t the answer
Some dead zones are not a Wi-Fi problem. They are a building problem, and the fix runs through a cable.
A detached garage. A basement under a poured slab. A room ringed in foil insulation. You can keep throwing radios at these rooms and keep getting the same answer. Run Ethernet, or run MoCA over the coax that is already in the wall, and put an access point on the far end of it.
And don’t wait for the ISP to solve it for you. Coverage past the gateway generally sits outside what the service actually covers, and the gateway’s location is usually constrained by where the service enters the home — not optimized for the farthest bedroom. Which is one more input into whether you keep the ISP’s gateway or buy your own.
The short version
A dead zone is not where your phone stops hearing the router. It is where the router stops hearing your phone. That is not how every weak link fails — but it is the failure your signal bars are most likely to hide.
So stop trusting the icon. Test the uplink against your own near-router baseline rather than an absolute number. Read the rate and the retries, wherever your router exposes them. And put the fix where the link is still good — not where it already failed.
Read next: the full slow-internet diagnostic checklist, why 6 GHz doesn’t reach the way you expect, and what channel width really costs you at the edge.
FAQ
Compare against yourself rather than against an absolute number. Run a speed test next to the router and write down both download and upload. Then run the same test, on the same device and server, in the room that feels bad. What you are looking for is upload falling away from your own near-router baseline, usually before download does. Run it twice a few minutes apart: a drop that repeats every time points to coverage, while wild swings more often point to interference, contention, or the test path itself.
Because bars only describe one direction. They show roughly how well your phone hears the router. They say nothing about how well the router hears your phone — and the phone is usually the weaker transmitter, so that direction tends to fail first. Every acknowledgment, upload, and outgoing side of a video call depends on it. The result is a link that looks healthy on screen and behaves like it is broken.
Not in the dead zone. Put it in the last place that still has a solid link back to the router. An extender cannot clean up a signal it did not receive cleanly, so placing it in the bad room simply rebroadcasts a bad connection. If the node cannot get a good link home, nothing behind it will either.
Often, but the backhaul decides how well. A repeater with no dedicated backhaul radio has to receive and retransmit on the same radio, and devices behind it can see roughly half the throughput. A dedicated backhaul radio avoids that specific penalty, but it does not make the wireless hop free. A wired backhaul — Ethernet, or MoCA over existing coax — removes the problem entirely, and for some rooms it is the only honest answer.
