Full Bars but Still Buffering: Signal Strength vs Real Speed

Four bars. The video still spins. Nothing about that is a contradiction, and it is the single most common Wi-Fi complaint there is.

Bars describe one thing: how loud the router sounds where you are standing. Real speed is the output of five numbers, and loudness is only the first. Once you can see the other four, the fix stops being a guess — and quite often the fix is not an extender at all.

Bars are a four-step summary of a sixty-point scale#

Indoor Wi-Fi signal spans roughly -30 dBm in the same room as the router down to about -90 dBm, where a connection stops existing. Your phone compresses that entire 60 dB range into three or four icon steps.

The consequence is blunt: on most phones anything better than about -65 dBm draws full bars. A spot at -35 dBm and a spot at -63 dBm show you the identical icon, and they differ by roughly 600 times in received power. Vendors also map the scale differently, so two phones held in the same doorway routinely disagree by a full bar.

So a full-bars room tells you one useful fact — you are probably above -65 dBm — and nothing else. What each dBm level actually supports is laid out in the dBm signal strength chart.

Five numbers, and only the last one is speed#

There is a chain between the router's radio and the file you are downloading. Each link is a gate, and a healthy value at one gate cannot rescue a bad value at the next.

Bars only ever report the first row. The second and third decide whether a loud signal is also a legible one, and a strong RSSI sitting in a noisy apartment band is the classic full-bars failure. As a rule of thumb, an SNR above 25 dB is comfortable, 20 dB is about the minimum for reliable throughput, and below 15 dB the link struggles however strong the RSSI looks.

Link rate is the number most people have never looked at, and it is the most diagnostic of the five. It is the modulation the two radios negotiated a moment ago: 6 Mbps at the bottom of the ladder, several hundred in a good room, over a gigabit next to a modern router. It moves constantly, and it is the honest answer to how good your connection is right now.

Throughput is the only one you feel.

Why the speed test never matches the link rate#

A link rate of 400 Mbps does not deliver 400 Mbps, and nothing is broken when it does not. Wi-Fi spends a large fraction of every second not carrying your data:

• It is half duplex. One radio talks at a time on a channel. Every frame your device sends stops the router from sending.
• Everything gets acknowledged. Frames, block acknowledgments, and the gaps between them all consume air.
• Devices back off before transmitting. Contention is how Wi-Fi avoids collisions, and it costs time whether or not anyone else is talking.
• Retries are invisible. A marginal link can retransmit a quarter of its frames and still report a healthy rate.

The working rule of thumb: expect 40 to 60 percent of the link rate as real throughput. Halving it in your head is close enough.

Airtime: how one slow device taxes the whole house#

This is the mechanism nobody is told about, and it explains a startling number of "the Wi-Fi is just bad" households.

Classic Wi-Fi shares transmit opportunities, not bytes and not seconds. Every device that wants to talk queues for a turn, and each turn lasts as long as that device needs. A device linked at 6 Mbps holds the channel for the entire duration of its slow transmission, and every other device — including the laptop three feet from the router — waits.

Put numbers on it. Sending one megabyte takes:

• about 20 milliseconds of air on a 400 Mbps link
• about a second and a third of air at a 6 Mbps legacy rate

Same megabyte, roughly sixty-five times the airtime. A 2.4 GHz-only camera parked in the garage at -80 dBm, an old printer, a decade-old tablet, a smart plug at the edge of the yard — any one of them can spend more of your channel than the rest of the house combined.

Many routers made in the last several years implement airtime fairness, which allocates equal time rather than equal turns. It limits the damage; it does not remove it, because the slow device still spends its share inefficiently.

What to do about it, in order of effort:

1. Open the router's client list and sort by link rate. Anything sitting under 20 Mbps is your suspect.
2. Move that device, or move it onto a band it handles well — which band reaches which room covers that choice.
3. Wire the ones that never move. A TV or a desktop on Ethernet leaves the air entirely.
4. If your router lets you, disable the oldest 802.11b data rates (1, 2, 5.5 and 11 Mbps).

The five-minute check#

Five steps, in order. Do them standing in the room that annoys you, with the device that annoys you.

1. Read the link rate, not the bars. On macOS, hold Option and click the Wi-Fi icon. On Windows, run netsh wlan show interfaces and read Receive rate. On Android, open the connected network's details. On iPhone, open the router's admin page and find the phone in the client list — iOS does not hand this to apps.
2. Run one speed test where you are standing. Note the download figure and the time of day.
3. Compare the two. Divide the speed test by the link rate. Above about 40 percent is normal. Far below it means airtime, not coverage.
4. Repeat both readings beside the router. If the link rate roughly doubles there, you have a coverage gradient. If the link rate is already high in the bad room, coverage is not your problem.
5. Plug a laptop into the router with a cable and test again. A slow wired result ends the investigation: the bottleneck is upstream of every radio in the building.

For the longer version of this — ten steps that also separate roaming, device faults and evening congestion — see the one-slow-room walkthrough.

When full bars mean an extender is the wrong purchase#

Say it plainly: if the failing room already reads better than about -60 dBm, an extender cannot fix it. An extender's entire function is to raise signal level in a place that lacks it. Where signal is not what is lacking, the box adds a second radio competing for the same airtime and makes the room slightly worse.

The four full-bars causes and what each actually needs:

• Channel congestion. Neighbors on your channel take airtime you can never buy back with more transmitters. Change channel or width first — see channels, neighbors and noise.
• A slow device dragging the household. Wire it, move it, or retire it.
• The client picked the wrong band. A phone clinging to 5 GHz through two walls will read weaker than 2.4 GHz would have on the same path.
• The bottleneck is upstream. A 50 Mbps plan is a 50 Mbps plan in every room in the house.

Only when the level in the room is genuinely low does placement become the lever, and then the question is where — which is what the halfway rule and its exceptions is for. A true coverage gap and a full-bars slowdown look identical from the couch; what a dead zone actually is separates the pair. The same pattern on a television has its own walkthrough in why a smart TV buffers on a strong signal.

Ruling coverage in or out on a floor plan#

The expensive mistake is buying hardware before you know whether signal level was ever the constraint. Range Up answers that half of the question without any hardware: you sketch your floor plan on the phone, pick the material preset your house is built to, put the router where it really stands, and read a predicted room-by-room coverage map.

If the map shows the problem room sitting comfortably in the green, you have eliminated coverage for free, and the five-minute check above points at the real culprit instead. If it shows a genuine hole, name that room and the app will work through possible extender spots, put them in order of how much each one recovers, and draw the leader against your current map. Nothing in it watches your live network — link rate, airtime and retries stay yours to measure — which is precisely why it can answer a question about hardware you have not bought yet. Try it on your own plan before the box arrives, or read how a second radio gets positioned first.