When somebody rings me and says the WiFi "doesn't reach the back bedroom", they almost always start by blaming the router. Nine times out of ten the router is fine. The problem is what sits between the router and the bedroom — and the honest answer is rarely the answer they want. So this is the practical, no-marketing version of what blocks WiFi signal in a London home, ranked roughly from a polite nudge to an outright wall of lead.

I have been working on phone lines, broadband and WiFi in London since 2011. In that time I have crawled through Victorian lath-and-plaster, modern foil-lined insulation, foot-thick brick chimney breasts, mansion-block concrete slabs, and one particular Aldgate steel lift shaft that should probably be in a textbook. The physics is the same everywhere. The materials are not. If you understand which is which in your own flat, you can stop buying gadgets that do not help.

How WiFi actually moves through a house

WiFi is just radio. It is the same thing as the radio in your kitchen, the same thing as a baby monitor, the same thing as the microwave on the counter — all of it is electromagnetic energy bouncing around at a particular frequency. Your router throws out a signal, the signal hits whatever is in the way, and some of it gets through. What does not get through is either absorbed (turned into a tiny amount of heat in the wall), reflected (bounced back the way it came), or diffracted (bent around the edge of an obstacle). The amount that survives is what your phone or laptop actually sees.

Signal loss is measured in decibels, written dB. The scale is logarithmic, which means it is brutal. Every 3 dB lost is half the power. Every 10 dB lost is a tenth of the power. So when I say a brick wall costs you 10 dB at 5 GHz, what I really mean is that the signal coming out the other side has roughly one tenth the strength it had on the way in. Add a second brick wall and you are down to one hundredth. The reason your router seems to drop off a cliff in the bedroom upstairs is not because it is broken — it is because dB loss adds up fast.

2.4 GHz versus 5 GHz versus 6 GHz — which one travels best

Modern routers transmit on at least two bands and increasingly three. Each behaves differently when it meets a wall:

  • 2.4 GHz is the old band. Long wavelength, decent at punching through brick and plaster, but slow and congested. In a London terrace it is fighting every neighbour's router, every Bluetooth speaker, every baby monitor, every cordless landline. Goes further, carries less.
  • 5 GHz is the working band. Far more spectrum, far less congestion, much higher throughput. Pays for it with worse penetration. Roughly speaking, every wall costs about twice as much signal at 5 GHz as it does at 2.4 GHz.
  • 6 GHz, added by 802.11ax (WiFi 6E) and 802.11be (WiFi 7), is clean spectrum and beautifully fast — but barely passes through a single plasterboard partition cleanly, never mind a load-bearing wall. Treat it as same-room only.

That is why your phone often shows a strong 2.4 GHz bar in the back bedroom and yet the connection still feels slow. You are getting the worst of both worlds: a band that travels well but is too congested to do anything useful, while the band that could actually carry your Netflix stream cannot get through the wall to reach you. For a deeper walk through the standards, I wrote a separate piece on WiFi 6 (802.11ax) that covers OFDMA, MU-MIMO and the rest of the alphabet soup.

The signal-loss league table, from polite to brutal

What follows is a rough working figure for each material at 5 GHz. Real-world losses depend on thickness, moisture content, framing and what is hidden inside, so treat these as ballpark numbers, not lab measurements.

  1. Open doorway: 0 dB. Air is free.
  2. Interior plasterboard partition (stud wall): 3 to 5 dB. Annoying but survivable. Most flats are mostly this.
  3. Wooden door: 3 to 4 dB. Same neighbourhood as plasterboard.
  4. Single-glazed window: 2 to 4 dB. Glass is mostly transparent to radio.
  5. Modern double or triple glazing (low-E coated): 10 to 20 dB. The metallic oxide coating on energy-efficient glass behaves like a partial mirror.
  6. Single brick wall: 8 to 12 dB at 5 GHz, 4 to 6 dB at 2.4 GHz. Most London Victorian internal walls.
  7. Double brick or external wall: 15 to 25 dB. Bay window returns, chimney breasts, party walls.
  8. Reinforced concrete slab or ceiling: 20 to 30 dB. Mansion blocks, new builds, basement conversions.
  9. Large mirror: 15 to 25 dB. The silvering is essentially a metal sheet.
  10. Fish tank, full bath, hot water tank: 15 to 25 dB. Water absorbs 2.4 GHz with particular enthusiasm because that is exactly the frequency your microwave oven uses to heat soup.
  11. Foil-backed insulation board (Kingspan, Celotex with foil facing) or foil-backed plasterboard: 30 to 40 dB. This is the silent killer. Looks like an ordinary wall, behaves like a mirror.
  12. Steel-clad RSJ, structural steel column, lift shaft, lift car: 40+ dB. Effectively a wall the signal cannot get round, only past.

Stack a couple of those together and you can see why the back bedroom is dead. A router in the lounge, an internal brick wall, a hallway, a foil-backed loft hatch and another stud wall is comfortably 50 dB of loss. You started with a fine signal and ended up with nothing.

Three London jobs that explain the maths

Stockwell — the Edwardian flat with foil-lined kitchen walls

A first-floor Edwardian conversion off Clapham Road. The owners had refurbished the kitchen during lockdown and the contractor, being conscientious about heat loss, had lined the external kitchen wall with foil-backed insulation board. Good for the energy bill. Catastrophic for the WiFi. The router sat in the front bay window. The bathroom, the only other room with a tiled wall, was directly the other side of the kitchen. Result: a phone in the bath showed nothing, not even a 2.4 GHz bar.

I scanned it with a proper analyser. Signal at the router was a healthy -42 dBm. Signal at the bathroom door was -88 dBm. That is a drop of 46 dB through two stud walls and a foil-lined kitchen — and the foil was doing almost all of it. The cure was not a booster. The cure was a single ceiling-mounted access point in the hallway, fed by a length of Cat6 stapled along the cornice and painted over. The bathroom now reads -55 dBm. The foil is still there. The signal goes round it, not through it.

Hampstead — the lounge with floor-to-ceiling mirrors

A beautiful house off Hampstead High Street with one of those long, narrow lounges. The end wall had been clad in floor-to-ceiling antique mirror panels for the look of the room. The router lived in the study at the far end of the house. The owners could not understand why one end of the lounge was perfect and the other end — the end with the mirrors — was a dead zone, even though the mirrors were not between them and the router.

This is where reflection bites you. A mirror does not just block what is behind it. It creates a region in front of it where the direct signal and the reflected signal arrive out of phase, cancel each other out, and leave a WiFi shadow that moves as you do. Stand a foot left and you are fine. Stand a foot right and your laptop drops. It is the same physics as a noise-cancelling headphone, only by accident.

The diagnostic that found it was a heat-map survey — twenty minutes walking the floor with a tablet plotting signal strength every couple of feet. The map made the cancellation pattern obvious. The fix, again, was a wired access point — this time tucked above the picture rail at the mirror end, so the signal arriving at the sofa came from the same side as the mirrors, not the opposite side. Cancellation gone.

Aldgate — the office where the lift shaft killed the corner meeting room

A small creative studio in Aldgate, top floor of a 1990s office block. Open-plan main floor, one corner meeting room with floor-to-ceiling glazing on two sides. The router sat on the reception desk by the lifts. Everywhere on the floor was fine. The meeting room, twenty metres away in a straight line, was unusable for video calls.

Why? Because the straight line went directly past the lift shaft, and the lift shaft is a steel-lined concrete box. The signal had to either pass through the shaft (impossible) or bend around it (terrible). What looked like a short hop on the floor plan was, in radio terms, a much longer journey through a building's spine. The meeting room was permanently in the lift shaft's shadow.

The owners had already tried a plug-in extender on the desk nearest the meeting room — that is more or less always the wrong tool, for reasons I cover in detail in a separate post on why boosters and extenders disappoint. The real answer was a ceiling access point in the meeting room itself, fed by Cat6 from the comms cupboard, powered over the same cable using PoE. The lift shaft no longer matters because the radio source is now on the right side of it.

Electrical interference — the invisible side of the problem

Not everything that kills WiFi is structural. Plenty of it is something plugged in down the hall. The 2.4 GHz band is a famously noisy bit of spectrum, shared with a long list of household devices that were granted the same frequency back when nobody thought it mattered:

  • Microwave ovens. A microwave leaks a small amount of 2.4 GHz energy by design. When it is on, a WiFi connection that depends on 2.4 GHz can stutter or drop completely. The fact that the band you cook your soup at is also the band carrying your video call is one of the more unfortunate accidents of consumer electronics history.
  • Old cordless landline phones (DECT and the older 2.4 GHz models). Modern DECT sits at 1.8 GHz and is fine. Old cordless handsets from the early 2000s were 2.4 GHz and squat right on top of WiFi channels.
  • Baby monitors and cheap video doorbells. Many of them still use 2.4 GHz analogue or low-bandwidth digital that hammers a single channel constantly. The router cannot work around them.
  • Wireless mice and keyboards, Bluetooth speakers, smart bulbs. Bluetooth and Zigbee both share the 2.4 GHz band. One device is nothing. A flat full of them is a chorus.
  • The neighbour's router. In a London mansion block you can typically see twenty other networks from your sofa. They are all fighting for the same handful of channels, and 2.4 GHz only has three that do not overlap.

This is part of why I push people toward 5 GHz wherever a wall allows. There is more spectrum, fewer fighters, and the signal-to-noise ratio (SNR) you can hold is much higher. SNR matters more than raw signal strength: a -60 dBm signal in a quiet band beats a -55 dBm signal in a band screaming with interference, every time.

Why a wall measurement beats a speed test

Almost everyone who calls me has already run a speed test, often several. They will tell me the line is delivering 200 megabits in the lounge and 8 in the bedroom. That is useful — but only a little. A speed test tells you the symptom. It does not tell you why.

What I do instead is measure the actual radio environment in dBm and SNR, room by room, on both bands, at human height. That gives you four pieces of information a speed test never will:

  1. How much signal is genuinely reaching the room (the strength, in dBm).
  2. How much noise is in the same band (which determines whether the signal is usable).
  3. Which direction the strongest signal is coming from — sometimes that is not the direction you expect, because of reflections.
  4. Where the cliff edges are — the points in the floor plan where signal drops 15 or 20 dB in the space of a metre, which almost always points at a specific material.

A speed test that says "8 megabits in the bedroom" tells you the bedroom is bad. A wall-by-wall radio measurement tells you that the bedroom is bad because the signal is dropping 22 dB at the chimney breast and another 14 dB at the foil-lined airing cupboard — and that is what tells you where to put the access point, or where to run the Cat6.

What the fix usually looks like

Once you know what is in the way, the engineering becomes simple. There are really only three workable strategies:

  • Move the radio closer to the problem. Almost always this means a wired access point fed by Cat6, powered by PoE, ceiling-mounted on the right side of the offending wall. This is the engineering answer and it is what I do for most London homes.
  • Go round the obstacle, not through it. If a brick chimney breast or a steel beam sits in the way, take the cable up into the loft or down to the floor below, then back up the other side. Radio cannot bend round things gracefully. Copper can.
  • Use mesh only where you cannot run cable. Mesh has its place — listed buildings, period flats where chasing walls is not an option, rental properties. But mesh with wireless backhaul is always a compromise, for reasons I cover in the mesh versus wired access points piece.

What you very rarely want is a plug-in extender bought from a high-street shop and dropped in the dead spot. The reason is structural, not brand-specific, and it is worth its own article — here it is.

Things you can check before ringing anyone

Before you call out an engineer, there are a few things worth verifying yourself. None of them require any kit beyond your own phone:

  1. Walk slowly from the router to the dead spot, watching the signal bars. Note the point where they drop sharply — that is your worst wall.
  2. Open the loft hatch and put the phone next to it. If the signal is dramatically better up there, the issue is foil-backed loft insulation reflecting your signal back down into the room below.
  3. Check the position of the router. If it is in a TV cabinet, behind a mirror, on top of a microwave or pressed against a foil-lined wall, move it first. Sometimes that is the whole job.
  4. Look at the back of the router. Disable 2.4 GHz on devices that do not need it (smart bulbs, doorbells) by giving them a separate guest network. Sounds small. Makes a real difference to SNR.
  5. If you have a phone with a "WiFi Analyzer" app installed, scan the rooms and note the channel congestion. If your channel and the neighbours' channels all overlap, switch.

If none of that shifts it, the building is doing more than the router can fight, and that is the point at which someone needs to come in with a proper analyser, sketch a heat map of the floor and tell you, definitively, what is in the way. For most London houses and flats that takes about an hour. Once you know what the obstacle is, the fix almost always involves a small length of Cat6 and an access point in the right ceiling — see the broadband engineer service page or the WiFi installation page for what that looks like in practice. I also cover the common patterns I see across London households in this round-up of recurring WiFi problems.

One last thing on the marketing

You will see a lot of routers advertised with frankly silly numbers — "AX6000", "AX11000", "tri-band gigabit" — implying coverage that will sail through anything. The numbers refer to theoretical aggregate throughput across all bands simultaneously to multiple clients using MIMO. They do not refer to how far the signal goes through your particular wall. No amount of marketing megahertz changes the fact that 5 GHz is going to lose 30 dB to a foil-backed insulation board, and no amount of MU-MIMO spatial streaming will get a usable connection into a room sitting in a lift shaft's shadow. The walls win. Always.

What changes the answer is moving the source of the signal to the right side of the walls. That is the whole engineering trick. Everything else is decoration.

Ring 020 3633 1131. Honest advice, freely given.