Blog · · Planning

How many WiFi access points do I need?

The short answer: for a typical home, one access point per 60–80 m² (650–860 sq ft) of drywall-divided space, one per 40–55 m² (430–590 sq ft) of brick, and one per 30–40 m² (320–430 sq ft) of concrete — at 5 GHz. Floor area on its own does not answer the question. Interior wall material moves the answer by a factor of two.

The table, by area and material

Read the row for your floor area and the column for what your interior walls are made of. Single storey; add roughly one access point per additional floor, and see the note on floors below.

Floor area Metric Drywall Brick Concrete
up to 800 sq ft up to 75 m² 1–2 APs 2 APs 2–3 APs
800 – 1,200 sq ft 75 – 110 m² 2 APs 2–3 APs 3–4 APs
1,200 – 1,700 sq ft 110 – 160 m² 2–3 APs 3–4 APs 4–6 APs
1,700 – 2,500 sq ft 160 – 230 m² 3–4 APs 5–6 APs 6–8 APs

These cells are not hand-picked. Each one is the top of that size band divided by the per-access-point area quoted above, rounded up — the arithmetic is in the source of this page, and it deliberately answers for the largest home in the row. Treat it as a starting guess to validate, not a specification. Above 2,500 sq ft the heuristic stops being useful, because at that size the shape of the building matters more than its area and you need the actual layout.

Why "one AP per 1,000 sq ft" keeps being wrong

Because it encodes no information about the building. The dominant term in indoor WiFi coverage is not distance, it is what the signal has to pass through. These are the attenuation figures our solver uses for a single crossing of one wall, following ITU-R P.2040 material properties at typical construction thicknesses:

Wall material 2.4 GHz 5 GHz 6 GHz
Drywall 3 dB 4 dB 5 dB
Solid wood 4 dB 6 dB 7 dB
Plain glass 2 dB 3 dB 4 dB
Brick 8 dB 11 dB 13 dB
Concrete 15 dB 20 dB 23 dB
Metal 25 dB 30 dB 32 dB

At 5 GHz a brick wall costs 11 dB and a concrete wall 20 dB, against 4 dB for drywall. Three brick walls between the access point and the far bedroom throw away 33 dB — more than the entire margin most people budget for. That is the whole reason the same square footage can need one access point or four, and the reason a rule of thumb keyed on area alone cannot tell you which.

Note the direction of the frequency columns. Attenuation rises with frequency, so a WiFi 6E or WiFi 7 network using 6 GHz needs more access points than the same layout on 5 GHz, which needs more than on 2.4 GHz. If you are buying 6 GHz hardware, plan on the 6 GHz column, not the friendly one.

The link budget, worked out

Here is the arithmetic behind a single point on a heatmap, so you can check ours. Take a typical access point at 20 dBm transmit with a 3 dBi antenna, and a laptop 10 m away on 5 GHz.

free-space loss at 10 m, 5000 MHz = 66.4 dB
line of sight: 20 + 3 − 66.4 = -43.4 dBm
through 2 brick walls: -43.4 − 2 × 11 = -65.4 dBm
through 2 concrete walls: -43.4 − 2 × 20 = -83.4 dBm

Two brick walls still leaves -65.4 dBm, which is comfortable. Two concrete walls leaves -83.4 dBm, which is the difference between a working video call and a spinning wheel — from the same access point at the same distance. Run that calculation for every point on a floor plan and for every wall on the path between it and each access point, and you have a predictive heatmap. That is exactly what the COST-231 multi-wall model does, and it is what the editor runs in your browser.

Every figure in this section is computed at build time by calling the same functions the heatmap calls. If the model changes, this page changes with it.

Floors count too, and usually get forgotten

A ceiling is a wall you have to cross vertically. A timber floor is cheap to pass through; a reinforced-concrete slab is close to opaque. The common failure is putting one access point in the middle of the ground floor of a three-storey house and assuming the top floor is covered — it is not, and no amount of transmit power fixes it. As a starting point, plan at least one access point per occupied floor and let the model tell you whether the middle one can serve two.

The other frequent mistake is turning transmit power up "to be safe". Your phone still transmits at around 15 dBm. Raising the access point to 23 dBm makes it shout further than it can hear, which produces rooms where a device shows full bars and cannot hold a connection. Leave it at 17–20 dBm and add an access point instead.

Decide it before you buy, in about 15 minutes

The table above gets you to "probably two, maybe three". The layout decides the rest, and the only way to settle it is to put the walls and the candidate positions on a plan. The sequence:

  1. Get a floor plan — an estate-agent PDF, a screenshot, or a photo of a sketch is fine. Predictive RF is ±6–10 dB anyway; half a metre of drawing error does not change the answer.
  2. Calibrate it with one known length. An interior door is about 80 cm in the EU, 32 inches in the US.
  3. Trace the interior walls and tag each one with its material. This is the step that does the work, and the step people skip.
  4. Drop the access points you are considering buying — pick the real model, so the transmit power and antenna gain come from its datasheet rather than a guess.
  5. Read the heatmap in the rooms you actually use. Aim for −67 dBm or better where you work and stream; the spare bedroom can be worse.
  6. Try one fewer. If the plan still holds up, you just saved the price of an access point.

For a full walkthrough with a real layout, see planning a 3-bedroom apartment. For which planning tools are genuinely free, see free WiFi planning tools compared.

What this will not tell you

Stated plainly, because a model you cannot see the edges of is not one you should trust. A predictive count does not account for furniture, multipath, co-channel interference from your neighbours, or throughput in Mbps — it answers signal strength, which is the term that decides access point count, and not the other terms that decide how the network feels once it is loaded. Expect ±6–10 dB against a real measurement, and do a ten-minute walk-around with a phone scanner after installation. If reality is within 8 dB of the prediction, the placement was right. If a specific spot is 15 dB out, something is there you did not draw — usually a metal door, a mirror, or an appliance.

Questions people actually ask

How many WiFi access points do I need for a 2,000 sq ft house?

Two if the interior walls are drywall, three if they are brick, four or more if they are concrete. Floor area alone does not decide it — a 2,000 sq ft (185 m²) house built from drywall needs roughly half the access points of the same house built from concrete, because one concrete wall costs about 20 dB at 5 GHz against a drywall wall's 4 dB. Count your walls and their material, not just your square footage.

Is one access point per 1,000 square feet a good rule?

Only for drywall single-storey layouts, which is the case the rule was written for. It carries no information about construction, and construction is the dominant term: at 5 GHz a brick wall attenuates 11 dB and a concrete wall 20 dB, against 4 dB for drywall. A house with the same footprint can legitimately need one access point or four.

Do I need more access points for WiFi 6E or 7 on 6 GHz?

Usually yes. Attenuation rises with frequency, so every wall costs more at 6 GHz than at 5 GHz and more at 5 GHz than at 2.4 GHz — brick goes 8 → 11 → 13 dB across the three bands. A layout that is comfortably covered on 2.4 GHz can have 6 GHz dead rooms with the same hardware in the same positions. Plan the band you actually intend to use.

Can I work out access point placement before buying the hardware?

Yes, and it is much cheaper than working it out afterwards. Predictive planning takes a floor plan, the wall materials, and the transmit power and antenna gain from the datasheet of the access point you are considering, then computes coverage from physics. WiFi Heatmap (wifiheatmap.app) does this in the browser with no download and no account, for any vendor, so you can compare "two of these" against "three of those" before spending anything.

How accurate is a predicted access point count?

The prediction lands within roughly ±6–10 dB of a real measurement in a typical home or small office. That is accurate enough to settle the question it is asked — two access points or three, and roughly where — and not accurate enough to replace a walk-around survey after installation. Furniture, multipath, and your neighbours’ networks are not modelled.

Check your own number — free, no signup

Upload your floor plan, trace the walls by material, and drop the access points you are thinking of buying. The heatmap updates as you move them. It runs entirely in your browser and the plan never leaves it. The free editor places as many access points as you like on one floor, and its export includes five — more than any size in the table above needs.