Why Your Wi-Fi Is Slow, and Why It Keeps Dropping
Slow Wi-Fi is almost never the internet connection, and Wi-Fi that keeps disconnecting is usually a different fault entirely. Here is how to tell which one you have, and what actually fixes each.
Two complaints arrive constantly, they sound similar, and they have almost nothing in common.
The first is the Wi-Fi is slow. The second is the Wi-Fi keeps dropping. Slow is a capacity problem. Dropping is a connection problem. Advice aimed at one routinely makes no difference to the other, which is why so much troubleshooting goes nowhere.
Here is what causes each, starting with the version we hear most in Greater Boston offices — because it comes with a clue built in.
The clue is the clock: it was fine this morning
Nobody changed anything. The router is the same router. The internet plan is the same plan. But at some point after lunch, video calls start stuttering, files take a minute to open, and someone walks over to ask whether the internet is down.
It is almost never the internet. Four things change between 8am and 2pm.
Everyone arrived. At 8am you might have six people on the network. By 10am you have thirty, each with a laptop and a phone, and the phone is on Wi-Fi whether the person is using it or not. Wi-Fi is a shared medium — every device takes a turn on the air, and the turns get shorter as the crowd grows. This is not a bandwidth problem, it is an airtime problem, and buying more internet does nothing for it.
Your neighbours arrived too. This is the part people miss, and it matters more in Boston than in a suburban office park. If you are in a multi-tenant building — Financial District tower, converted mill in Waltham, upper floor above a storefront in Quincy Center — every tenant is broadcasting on the same handful of channels. At 8am half of them are empty. By 2pm the airspace is saturated with traffic that is not yours and that you cannot control. Your equipment is fine; the room it is shouting in got loud.
Something started running. Cloud backup, OneDrive sync after a big file drop, Windows updates, a Zoom recording uploading. Plenty of these default to business hours. One machine pulling a large sync can consume most of an access point’s capacity while contributing almost nothing visible.
The microwave. Not a joke. A microwave oven leaks at 2.4 GHz, which is the same band older devices and most cheap access points prefer. Offices where the kitchen is near the access point get a reproducible slowdown between noon and one o’clock, and it looks exactly like a mystery until someone maps it to lunch.
The cause that surprises people: one slow device
Wi-Fi negotiates a speed with each client separately. When a device connects at a low rate — an old printer, a warehouse scanner, a laptop from 2016 clinging to 2.4 GHz from across the floor — it holds the air for far longer to send the same amount of data.
The effect is not proportional. A single legacy device transmitting slowly can consume a disproportionate share of an access point’s airtime, and everyone else waits. Modern access points have airtime fairness features that limit the damage, but they have to be turned on and configured, and on equipment bought at a big-box store they frequently are not.
This is the most common single fix we apply: find the device dragging the access point down, move it to a wire or a separate radio, and the whole floor gets faster without a dollar of new hardware.
When it is not slow, it is dropping
Disconnections have their own set of causes, and the first thing to establish is whether everyone drops at the same moment or one person drops repeatedly. That single question splits the list in half.
Everyone at once, briefly, for no obvious reason. On 5 GHz, a large block of channels is shared with radar under rules called DFS. When an access point detects a radar pattern on one of those channels, it is required to vacate immediately and stay off it — no grace period, no warning. Every device on that radio drops and has to reconnect. Near Logan, along the coast, and anywhere in range of weather radar, this is a real and reproducible cause that looks completely random from the inside. The fix is to pin the access points to non-DFS channels and accept a smaller pool.
Everyone, around the same number of devices, every time. Small routers ship with a small pool of addresses to hand out, and thirty staff routinely means seventy devices once phones, tablets, printers, thermostats and the conference room TV are counted. When the pool runs out, new devices cannot join and existing ones fail to renew. It presents as random, it is arithmetic.
One person, repeatedly, while moving. Their laptop is holding on to an access point it walked away from instead of switching to the near one. Devices decide when to hand over, not the network, and some are stubborn about it. Access points that are too powerful make this worse, because the distant one still sounds loud enough to keep.
One person, repeatedly, sitting still. Usually power management on the network adapter putting the radio to sleep, or a driver that has not been updated since the machine was bought. This is a device fix, not a network fix, and chasing it at the router wastes an afternoon.
Calls drop but files are fine. Voice and video are unforgiving about brief gaps that a file transfer simply retries through. A network can be entirely adequate for documents and visibly bad for Teams calls, and both observations are correct.
Why Boston buildings make both problems worse
We work in a lot of nineteenth-century brick with plaster and metal lath, and in converted industrial space with structural steel and high ceilings. Both are hostile to wireless in ways that modern drywall construction is not.
Plaster over metal lath behaves close to a mirror for Wi-Fi. Signal that should reach the back office instead reflects, and what does arrive is weak enough that devices drop to lower rates — which brings back the airtime problem above, and puts devices close enough to the edge that they disconnect. In converted mill and arsenal buildings, the ceiling height means an access point mounted high covers a wide floor area weakly rather than a sensible area well.
The instinct is to buy a stronger access point. That usually makes it worse: a louder transmitter reaches further into your neighbour’s space and theirs into yours, the client devices still cannot shout back at the same volume, and stubborn laptops cling to it from further away. The answer is almost always more access points at lower power, placed where people actually sit.
How to diagnose it yourself in ten minutes
You can get most of the way with no tools:
- Run a speed test on a cable. Plug a laptop into the router or a switch port. If wired is fast while Wi-Fi is slow at the same moment, the internet is not your problem and no upgrade will help.
- Test in two places. Next to the access point and at the worst desk. A large gap means coverage. Similar bad numbers everywhere means congestion or backhaul.
- Watch the clock for a week. Note the time whenever someone complains. A pattern that tracks the lunch hour, or the start of the workday, or a specific recurring meeting, narrows the cause faster than any measurement.
- Count the devices. Log into the router and look at the client list. People are routinely surprised — thirty staff often means seventy connected devices once phones, tablets, printers, thermostats and a smart TV in the conference room are counted.
- For drops, ask who and when. One person or everyone; moving or sitting still; calls only or everything. Those three answers usually name the cause before anyone touches the equipment.
If wired is fast, coverage is even, and the problem still tracks the clock, you have congestion. That is a design problem, and it is fixable — but not by replacing the router with a slightly newer router.
What a real fix looks like
For most small offices it is a short list, in this order: measure what is actually happening, plan channels so you stop fighting your neighbours and yourself, place access points for the desks that exist rather than the geometric centre of the floor, separate guest traffic so visitors cannot consume the capacity your staff needs, and get the chronically slow devices onto a wire.
Hardware comes last, and often does not come at all.
If your Wi-Fi has a daily pattern nobody has explained, or drops at moments nobody can predict, text us and describe when it happens. Half the time we can name the likely cause from the timing alone.
Frequently asked questions
- Will a faster internet plan fix slow Wi-Fi?
- Almost never. Internet speed is the pipe leaving your building; Wi-Fi is how traffic reaches that pipe from each desk. If a speed test run on a cable is fast while the same test on Wi-Fi is slow, more internet buys you nothing. We check the wired speed first for exactly this reason — it settles the question in two minutes and saves people from upgrading a plan that was never the problem.
- Why does my Wi-Fi keep disconnecting rather than just being slow?
- Dropping and slowing are different faults with different causes, which is why the usual advice fails on one of them. Slowness is normally congestion or a weak signal. Dropping is normally your device being handed between access points badly, running out of addresses to lease, going to sleep to save battery, or the access point being forced off its channel. The useful first question is whether everyone drops at the same moment or just one person drops repeatedly — the first points at the network, the second at the device.
- How do I tell interference from congestion?
- Interference is worst near the source and often affects one area or one band. Congestion follows the number of people online and affects everyone at once. A rough test: if the problem tracks the clock — bad at 10am and 2pm, fine at 8am and 6pm — it is congestion or backhaul. If it tracks the place — always bad in the back conference room — it is coverage or interference.
- We have mesh units from a big-box store. Is that the problem?
- Sometimes, and not for the reason people expect. Consumer mesh works well in a house with a dozen devices. An office with thirty clients, a printer, a couple of access points from different generations and a neighbouring tenant broadcasting through the wall is a different physics problem. The gear is not fake, it is just built for a load you have outgrown.
- How long does a Wi-Fi survey take?
- A small office is usually a couple of hours on site plus a written result: a floor plan marked with signal strength where people actually sit, the channels in use around you, and the fixes ranked by cost. We do the survey before recommending hardware, because most of the offices we walk into need placement and channel planning rather than new equipment.