Most weeks I get called to a house where the customer has already tried two or three plug-in WiFi extenders before picking up the phone. Sometimes more. The boxes are usually still in a kitchen drawer, the little green lights still cheerfully glowing, the WiFi still not working in the room they were supposed to fix. So when somebody asks me, do WiFi boosters work — the honest answer is yes, technically, in a very narrow set of circumstances, badly. And in most London homes, those circumstances do not apply.

I have nothing against the manufacturers. The devices do what they say on the box. The trouble is that what is written on the box is not the same as what the customer hears in the shop, and what the customer hears in the shop is not the same as what the physics will deliver in a Victorian terrace with a chimney breast in the middle of the floor plan. This article is the version I give people on the phone before they spend any more money on the wrong tool.

What a "booster" actually is, mechanically

Let us be precise about the terms because the high street is wonderfully sloppy with them. The boxes you see in Currys, Argos, Robert Dyas and on Amazon under names like "WiFi Booster", "WiFi Extender", "Range Extender" and "WiFi Repeater" are, almost without exception, the same device with different branding. The device is a small radio that does two jobs at once:

  1. It listens to your existing router's WiFi signal.
  2. It re-broadcasts that same signal under either the same name or a slightly different one.

That is it. There is no second cable. There is no separate connection to the broadband. The only path the extender has back to the rest of the internet is the very same WiFi signal that was not strong enough in the first place. Read that sentence again — it is the entire reason these devices disappoint.

The signal-halving problem

Here is the bit that nobody at the till tells you. When an extender sits between your phone and your router, the wireless connection has to make two hops instead of one. Hop one: router talks to extender. Hop two: extender talks to your phone. Both hops have to share the same radio. The extender cannot listen and talk at the same time on the same frequency, so it has to take turns. In practice, your effective throughput is roughly halved.

Worse, the extender can only re-broadcast a copy of the signal it is hearing. So if the extender itself is sitting in a spot where the router's signal is already weak — and it usually is, because that is the only place it makes sense to plug it in — then everything past it is downstream of a weak link. You have not boosted the signal. You have made a copy of a tired signal and given it a louder voice. Volume is not the same as clarity.

The way I sometimes put it to customers is this: if your router is whispering by the time it reaches the hallway, an extender in the hallway is a man with a megaphone shouting the whisper. You can hear him better, but he has nothing useful to say.

Why "70% signal" is misleading

One of the most common things I hear is "but the extender app says I am at 70% signal strength, that should be plenty". It is not plenty. The bars in those apps are largely cosmetic. They measure raw received signal strength in dBm — a single number — and then map it to a percentage with whatever scale the developer fancied. They do not measure:

  • SNR (signal-to-noise ratio). The amount of signal above the background noise. In a London mansion block you can have a strong signal and still be drowned by every neighbour's router, baby monitor and Bluetooth speaker on the same channel. SNR is what actually determines whether the connection works.
  • Retransmission rate. The percentage of packets that have to be sent twice because they got corrupted in flight. Modern WiFi (802.11n, ac, ax) leans heavily on getting the packet through on the first try. Once retransmissions creep above about 10%, the throughput collapses, even though the signal "bars" still look fine.
  • Modulation rate. A clean link uses high-order modulation (1024-QAM on ax) and pushes a lot of data per symbol. A degraded link falls back to lower-order schemes — sometimes all the way down to 1 or 2 megabits — and the bar graph never tells you it happened.
  • Round-trip latency. Two-hop wireless adds milliseconds and jitter. Web pages load eventually, but video calls fall apart.

This is why an extender can look "fine" on its dashboard and still feel unusable in the next room. The dashboard is not measuring the things that matter, and the budget the device was built to is not going to put the right kit in the box to measure them properly. I cover the gap between consumer signal bars and a genuine radio survey in the post on heat-map surveys.

The dual-band exception (and why it is still a compromise)

There is one class of extender that is less terrible: the dual-band model that uses one band (usually 5 GHz) exclusively for the backhaul link to the router, and the other band (often 2.4 GHz) for talking to the clients in the room. This is sometimes called "dedicated backhaul". Because the two hops happen on different frequencies, the device can listen and talk at the same time, and you do not lose half your throughput.

That helps. It does not save you. The 5 GHz backhaul still has to make it from the router to the extender through whatever walls are in the way — and 5 GHz, as I cover in the piece on what blocks WiFi signal, loses roughly twice as much through every wall as 2.4 GHz does. So the band carrying the backhaul is the one that copes least well with the building. You are still throwing radio at the problem instead of cable, and the problem is still mostly a building.

Tri-band mesh kits — which dedicate an entire third radio purely to backhaul — are a real improvement, and there is a place for them in flats where running cable simply is not possible. But once you are at that tier of kit you are well past the budget plug-in extender that started this article, and you are within sight of doing it properly with a wired access point anyway.

Consumer kit versus prosumer kit — the gap is wider than you think

The thing that genuinely surprises customers when I open my van and show them what I actually install is how much bigger and heavier a proper access point is than a plug-in extender. A high-street "AC1200" extender weighs about as much as a cupcake. A prosumer ceiling access point from the kind of brand you would not have heard of unless you have run a small office network — Ubiquiti, TP-Link Omada, Aruba Instant On, Engenius, the like — weighs the better part of half a kilo, has antennas the size of a thumb hidden inside the dome, transmits at the full legal limit, runs MU-MIMO with two or three spatial streams on each band, supports 802.11ax (WiFi 6) with OFDMA properly, and is designed to be powered by Cat6 from a PoE switch in a cupboard.

The differences that actually matter:

  • Transmit power. A consumer extender often runs well below the legal maximum to keep heat and cost down. A proper access point runs at the legal ceiling.
  • Antenna design. Cheap kit has tiny PCB antennas. Real kit has properly tuned arrays, often with beamforming that aims the radio at the device that is talking.
  • Backhaul. Wired, Cat6, no shared radio, no halved throughput, no dependence on the wall being thin.
  • Roaming. Proper access points support 802.11k/v/r so your phone hands off cleanly between APs as you walk through the house. Extenders typically do not, which is why people complain that the WiFi "sticks to the wrong one" as they move around.
  • Channel management. A controller-based setup notices channel congestion and shifts the radios automatically. An extender will happily sit on a broken channel forever.

This is not really a brand argument. It is a category argument. A high-street plug-in extender is not a smaller, cheaper access point. It is a different category of device that is doing a different job, badly, because that is what its hardware was built for. The full comparison sits in the mesh versus wired access points piece.

Three London jobs where extenders failed first

Greenwich — four extenders and a frustrated family

A semi-detached in Greenwich, three storeys, family of five, router in the lounge on the ground floor, two teenagers gaming in the loft bedroom. The owner had bought, in order: a plug-in extender from the supermarket, a slightly more expensive extender from Argos, a third "long-range" model marketed as "for large houses", and finally a "mesh" two-pack from a high-street brand. He showed me all four when I arrived. The first three had failed within a week. The mesh kit was the only one still running, and it was dropping out under load every evening at about 7 pm.

When I measured the house, the reason became obvious. The router was sitting in a media cabinet at the front of the lounge. The chimney breast and a foil-backed insulation board in the loft conversion were directly between the router and the loft. Every extender he had tried needed to plug in somewhere between the two, and there was no useful socket between them where the signal was still strong enough to repeat usefully. The mesh kit was doing the best it could but its wireless backhaul was struggling with the same wall the extenders had struggled with.

I ran a length of Cat6 from the router up the back of the chimney breast, into the loft eaves, across the joists and into a ceiling-mount access point in the middle of the loft hallway. PoE down the same cable. One AP, one wire, one job. The teenagers' ping fell from 60-plus milliseconds with jitter to under 5 milliseconds steady. The four extenders went into a drawer. The family stopped fighting about Fortnite. Total kit list: one cable, one AP, one PoE injector. The cost of doing it properly was less than the four wrong things he had already bought.

Maida Vale — when powerline adapters drop out as the lights are switched

Powerline adapters are the close cousin of the WiFi extender — instead of using radio, they use your electrical wiring to carry the network signal between two plugs. People love them because they look like the perfect solution: no holes drilled, no cable run, just plug-and-go. Sometimes they really do work. Often they do not, and the failure modes are particularly maddening because they are intermittent.

This Maida Vale flat is a classic. Mansion block conversion, kitchen at one end, study at the other. The owner had two powerline adapters carrying his broadband from the router in the kitchen to a desktop in the study. It worked perfectly during the day. Every evening at dusk, the link would drop out. He had assumed it was the broadband, called the ISP, been told the line was fine, and lived with it for two months before ringing me.

The cause turned out to be the lighting. The flat had been refitted with LED downlights through the entire hallway. LED drivers — particularly cheaper, dimmable ones — pump electrical noise back onto the mains at exactly the frequencies powerline adapters use. When the hallway lights came on, the powerline link's signal-to-noise ratio collapsed and the connection died. During the day, with the lights off, it was clean. At night it was unusable.

Powerline also struggles in three other London-specific situations I see constantly:

  • Two different ring mains. If the kitchen and the study are on different circuits at the consumer unit, the powerline signal has to jump the bus bar inside the fuse board. Sometimes it does. Often it does not.
  • RCDs and surge protectors. Modern RCDs and any plug-in surge-protected extension lead will filter out the powerline carrier as if it were noise. Plug a powerline adapter into a surge strip and it will not work.
  • Old aluminium wiring or pre-war rubber-insulated wiring. Higher resistance, poorer connections, lots of attenuation. Powerline signal arrives at the far end too weak to decode.

The fix in Maida Vale was the same fix as Greenwich — a single run of Cat6 along a coving line, a ceiling AP at the study end, problem gone. The powerline adapters went the way of the extenders.

Soho — when the mesh kit from a high-street shop couldn't cope with floor 3

A small post-production studio occupying three floors above a Soho restaurant. Edit suites on the top floor, account managers and admin on the middle floor, kitchen and reception on the ground floor. The owner had bought a three-node mesh pack — well-reviewed brand, decent kit on paper — and dotted the units one per floor. Performance on the ground floor was fine. The middle floor was acceptable. The top floor, where the editors needed to push enormous video files, was dreadful.

I scanned the building. The ground-floor mesh unit was happily talking to the middle-floor unit through the 5 GHz backhaul. The middle-floor unit was struggling to reach the top floor through a concrete slab. Concrete eats 5 GHz alive — easily 25 dB of loss per floor in that kind of building — and the third hop through the second concrete slab meant the top-floor unit was getting a signal so weak that its own clients were sharing what little bandwidth remained over a wireless link that was already on its knees.

The mesh manufacturer had not lied. The kit was capable of mesh backhaul. The building, made of concrete poured in the 1960s, was simply not a building you could throw radio at and expect it to win. Three floors of slab demand wired backhaul. The answer was to take the existing mesh nodes, switch off their wireless backhaul, drop a run of Cat6 up the lift shaft riser to each floor and feed them by ethernet. Same kit, same SSIDs, no client-side change. The editors stopped complaining within the hour.

The right answer in most London homes

The honest engineering answer for almost every London property I see is the same: one or more wired access points, fed by Cat6, powered by PoE, mounted on the ceiling. Not in a TV cabinet. Not on a shelf. On the ceiling, where the radio is high enough to clear furniture and the antenna pattern has a fighting chance.

The reasons this beats a plug-in extender in essentially every circumstance:

  • No wireless backhaul. The data is carried by copper from the AP to the router. Walls cannot weaken copper.
  • Full transmit power. The AP runs at the legal limit because it is mains-fed via PoE rather than running on a tiny SMPS inside a 13 A plug-top.
  • Proper roaming. Devices hand off cleanly as people walk between rooms.
  • One SSID, one password, one network. No "_EXT" suffixes, no manual switching, no devices clinging on to the wrong radio.
  • Future-proof. Cat6 happily carries 1 Gbps and (with the right kit) 10 Gbps. The same cable will outlive several generations of WiFi standard.

The trick is not the AP — there are many that will do the job — it is putting it in the right place, knowing where to run the cable, and not chasing a wall when you do not have to. That is where having an actual engineer in the building, with a radio analyser and a tape measure, earns its keep. Have a look at the WiFi installation page for what a typical install involves, the ethernet cabling page for how cable gets run cleanly in occupied homes, and the broadband engineer page for the wider picture.

When a booster genuinely is the right tool

In fairness to the device, there are scenarios where a plug-in extender is fine. They are narrower than the marketing suggests, but they exist:

  1. One specific dead spot, one device, no video calls. A printer in the back of a garage. A smart bulb on a landing. A Ring doorbell at the front door. Things that need a low-bandwidth, latency-tolerant connection in one specific place.
  2. Open-plan flats with line-of-sight from router to extender to dead spot. If the path is mostly air with one stud wall, the device does what it says.
  3. Listed buildings where running cable is not legally an option. In which case a proper tri-band mesh kit with dedicated backhaul will outperform a plug-in extender by a margin.
  4. Rentals with hard limits on what you can install. Same caveat — go for proper tri-band mesh rather than the cheap high-street box.

Outside those cases, you will spend more money chasing the symptom with successive extenders than you would have spent fixing the cause the first time. I have seen drawers with four of them. I have seen one Hampstead house with seven. The boxes are very forgiving — they keep their little green lights on for years, even when nothing is going through them.

The short version

Do WiFi boosters work? In a tightly bounded set of scenarios, yes — a little, and reluctantly. In a typical London home with brick walls, foil-backed insulation, concrete slabs, mirrored alcoves and a chimney breast in the worst possible place, no — not in any way that will satisfy a video call or a games console. The physics are stacked against the device. A wired access point sidesteps the physics entirely by not relying on radio for the backhaul, and that is why it is what gets installed almost every time once someone has stood in the building with proper measuring kit. If you are already past the second or third extender in the drawer, you have my sympathy — and the next call is the one to make.

Ring 020 3633 1131. Honest advice, freely given.