Buy a router with a big number on the box — 3,000, 6,000, more — run a speed test the day it arrives, and the figure that comes back is a fraction of what the packaging promised. The instinct is that something must be broken: the router, the plan, the install. Usually nothing is broken. The gap between the box and the speed test is not a defect. It is the design showing through.
I spent years on the side of this industry that produces those numbers — qualifying radios and signing off on the rates that eventually get printed on the packaging. So here is the part that never makes the marketing: the number on the box is a best-case physical-layer link rate built from theoretical radio capabilities under ideal conditions. What reaches your laptop is what’s left after the protocol, the air, and your own device each take their cut.
Walk through those cuts and the gap stops being mysterious. This is one piece of a larger pattern — every Wi-Fi generation is sold as a bigger number, which is the throughline of the State of Wi-Fi overview — but nowhere is the gap between the number and the experience more obvious than here.
The box number is not your speed
The figure on the box is a PHY rate: a physical-layer link rate. It is how fast the radio can, in theory, push bits over the air under best-case conditions — strong signal, no interference, the widest channel, the most antennas, the densest modulation, all at the same time. It is a best-case ceiling, not a number any application was ever promised.
It is also, very often, a sum. An AX3000 router does not mean your laptop gets 3,000 Mbps. It usually means the router is adding together the best-case rates of different radios — for example, one number for 2.4 GHz and another for 5 GHz — even though a single client normally uses only the link, or links, it actually supports. The “3000” was never a single speed your laptop was promised; it is a stacked radio-class number for the label.
That is not dishonest, exactly. It is a spec. But a spec built from perfect conditions is the start of the story, not the speed you get. The box number is useful for comparing radio classes. It is not useful for predicting the speed of one phone, in one room, through two walls, on a Tuesday night.
The first cut: protocol overhead
Even in clean conditions, a large part of the PHY rate never becomes application throughput. Some of it is protocol overhead — the headers, acknowledgments, and management frames every Wi-Fi conversation needs just to stay organized. Some of it is airtime contention — devices taking turns, because they cannot all talk at once. Some of it is the simple fact that Wi-Fi is a shared, half-duplex medium: a radio can talk or listen, but not both at the same instant.
None of this is a malfunction. It is how shared wireless works. So when a real speed test lands far below the box number, that is not automatically a defect. It is usually the design showing through.
The second cut: your device
Your router may be 4×4. Your phone probably is not.
This is the cut almost no one accounts for. People read the router’s spec sheet and quietly assume their devices will use all of it. They won’t.
A high-end router might have four spatial streams, three bands, and support for the widest channels. The phone in your hand is often two spatial streams, may not have a 6 GHz radio at all, and may negotiate a narrower channel than the router can offer. A Wi-Fi link is only ever as capable as the narrower end of it — and the narrower end is almost always the client device, not the router. You can buy the most capable access point on the market, and your laptop from three years ago will still cap the connection at what it can do.
It is also why two phones standing in the same spot, on the same router, can test completely differently. Features like Multi-Link Operation only help if the client supports them too; the box number assumes a perfect partner device that most people do not own.
The third cut: the air
Wi-Fi does not hold a fixed speed. It negotiates one, continuously, based on conditions. Move away from the router, put a wall in between, let a neighbor’s network land on the same channel, and the radios quietly step down to a slower, more robust modulation to keep the link from breaking.
That step-down is a good thing — it is what keeps the connection alive instead of dropping — but every step down is a slower link. Distance, walls, and interference are not edge cases; they are the normal condition of a real home. The box number assumes none of them exist.
The fourth cut: everything after Wi-Fi
Even if the wireless link were flawless, it is only one segment of a longer path — and any segment can be the bottleneck.
Your internet plan sets a hard ceiling no router can beat: a 500 Mbps plan will never speed-test at gigabit, no matter what the box says. The router’s WAN port and your device’s Ethernet port set their own limits. And the server on the far end has a speed too — a slow download often says more about the source than your Wi-Fi.
Mesh adds one more. A mesh node without wired backhaul has to spend airtime talking to your device and more airtime talking back to the main router. That does not make mesh bad. It just means the backhaul is part of the speed test — and a wireless hop can cut throughput substantially on its way to the far node.
What number should you actually expect?
So what is normal? As a rough rule of thumb, real-world throughput often lands well under half the headline figure — and that is before distance and walls take their share. A strong, close connection on a good router can feel genuinely fast and still test at a fraction of the box number. Both things are true at once.
The useful question is not “why am I not getting the number on the box” — you were never going to — but “is this consistent with my plan, my device, and where I’m standing?” If a speed test at close range comes in well below your paid plan, that is worth investigating. If it comes in below the box number, that is just Wi-Fi.
How to read a router box
Three things to hold onto next time you’re comparing numbers on a shelf:
- The headline figure is a best-case, lab-measured PHY rate, usually summed across bands — not a speed any single device receives.
- Your real speed is capped by the slowest link in the chain, which is frequently your own device, then the air, then your plan.
- A speed test below the box number is normal. A speed test below your paid plan, at close range, is the one worth chasing.
The box number is not a lie. It is just not the number your laptop was ever going to see.
FAQ
Because the number on the box is a best-case physical-layer link rate measured in lab conditions, not the throughput your device receives. Protocol overhead, the shared half-duplex nature of Wi-Fi, your own device’s limits, distance, and interference all reduce the real figure — so a speed test well below the box number is normal, not a fault.
The PHY rate (or link rate) is the theoretical speed the radios negotiate over the air. Real throughput is what actually moves your data after overhead, contention, and retransmissions are removed. Throughput is always lower than the PHY rate — often well under half of it, even on a strong connection.
Your plan and the router’s box number are two different ceilings, and your speed test is limited by the lowest one in the chain. If a close-range test comes in below your paid plan, the next step is to isolate the bottleneck: test wired if you can, check the router’s WAN and LAN port limits, and compare more than one test server. The cause may be Wi-Fi, your device, placement, a wired bottleneck, or the service path beyond your home — but it is a different question from the number printed on the router box.
Most often it’s the device. A router may support four streams, three bands, and wide channels, but a typical phone or laptop uses fewer streams and may not support the newest bands or channel widths. The connection is capped by the weaker end, which is usually the client — so upgrading the router alone often changes less than people expect.
