What is a Wi-Fi survey?
A plain-English reference. What a survey measures, how it is actually done, what the numbers mean, and what you get at the end — written so you can forward it to whoever is approving the spend.
The short answer
A Wi-Fi survey measures how radio behaves in a specific building and turns that into a design. Where the access points go, what they transmit at, which channels they use, how many devices each can carry, and what cabling has to reach them.
It exists because radio does not behave the way a floorplan suggests. A stud wall costs a few decibels; a lift shaft costs most of a signal; a warehouse full of steel racking behaves like a completely different building once the pallets are in. None of that is visible on a drawing, and all of it decides whether the network works.
The output is a document rather than an opinion — which is what lets you tender the work on equal terms, hand it to an installer and get a comparable price, and hold somebody to a standard when it goes in.
The two measurements that matter
Almost every conversation about Wi-Fi performance comes back to two numbers, and most people have only heard of the first.
Signal strength (RSSI) is how loud the access point is where you are standing, measured in dBm. It is negative and logarithmic: −45 dBm is excellent, −65 dBm is a sound working level, −80 dBm is nearly unusable.
Signal-to-noise ratio (SNR) is the gap between that signal and the background noise on the same channel, in dB. It is the better predictor of whether a connection actually works, because a strong signal in a noisy environment still performs badly — and it is the number a cheap survey quietly leaves out.
How a survey is actually carried out
Not a walk round with a laptop. This is the method our surveys are run to, using Ekahau Sidekick and Ekahau Site Survey Pro with a real access point as the reference.
Plan
We start on the drawings: floor area, construction, ceiling heights, and the number and type of devices that have to work. That produces a first view of how many access points are needed and roughly where.
Measure
A survey access point goes up on a tripod at the first candidate position. A surveyor walks the area around it recording real measurements, then that position is fixed in the model and the process repeats for the next one.
Model
The measurements from each position combine into a single picture of how the finished network would behave — coverage, signal-to-noise, channel use and roaming, per band.
Report
Coverage and SNR plots per floor and per band, access point positions with mounting photographs, what the neighbouring networks are doing, and a costed bill of materials.
The equipment matters because the measurements have to be calibrated and positioned accurately on a scaled drawing. A laptop's built-in radio reports what that particular laptop sees; a survey adapter with spectrum analysis reports what is actually happening in the band, including the sources that are not Wi-Fi networks at all and would never appear in any consumer tool.
The reference access point matters too. Modelling propagation from a datasheet is an estimate. Putting the real unit on a tripod at the exact position and height being considered, and walking the area, is a measurement.
What you get at the end
A survey report is a working specification, not a certificate. Ours is structured so a facilities or IT manager can understand what is wrong, a finance director can budget from it, and an installer can build from it without a clarification email.
- Introduction — objectives, requirements, scope and constraints, so everyone agrees what was and wasn't covered
- Test equipment and method — exactly how the measurements were taken
- Assessment considerations — the signal strength and signal-to-noise thresholds the design works to
- Access point placement — coverage and SNR plots per floor and per band, with each position marked
- Placement photographs — a picture of each proposed position and the comms room, so nothing is ambiguous on site
- Neighbouring networks — what else is transmitting around you, and what it costs you
- Recommended bill of materials — costed, with part numbers, so the work can be tendered
What can invalidate one
Radio propagation is a property of the building as it stands. A survey describes the building on the day it was measured, and these are the changes that move the answer:
- Walls and partitions — including a change of material, or wall covering added or removed
- Ceiling height changed, or a false ceiling installed
- Floor finishes changed — carpet or timber added or taken up
- Lighting changes, particularly dimmable fluorescent fittings
- Furniture, especially metal cupboards and shelving close to an antenna
- Large plants, or large numbers of small ones, real or artificial
- Anything over about 1.4 m tall placed near an access point
- Other wireless systems, or equipment sharing the same bands
- People — a change of roughly half the occupancy either way
None of that means a survey has a short shelf life. It means a material change to the building or the load is a reason to re-check the affected area — which is exactly what an annual RF health check is for.
The terms, in plain English
Everything you are likely to meet in a quotation, a report or a supplier conversation.
Access point (AP)
The unit on the ceiling or wall that devices connect to. Not a router — in a commercial network it is one of many radios sharing a design, fed by a data cable that also powers it.
RSSI
Received signal strength, in dBm — how loud the access point is where you are standing. The number most people mean by "signal". Necessary, but on its own the least informative measurement in a survey.
dBm
The unit signal strength is measured in. It is negative and logarithmic, so closer to zero is stronger: −45 dBm is excellent, −65 dBm is a sound working level, −80 dBm is nearly unusable. A 3 dB change is roughly double or half the power.
Signal-to-noise ratio (SNR)
The gap between the signal and the background noise on that channel, in dB. The best single predictor of whether a connection actually works. Strong signal in a noisy environment still performs badly.
Noise floor
Everything on the channel that is not the signal you want — neighbouring networks, machinery, and any other emitter in the band.
Co-channel interference
Two or more of your own access points on the same channel, in range of each other, taking turns on the same airtime. The most common self-inflicted fault in commercial Wi-Fi, and usually the result of adding access points without a channel plan.
Airtime
The finite resource every device on a channel shares. Capacity problems are airtime problems: forty devices on one radio queue for it, whatever the signal strength says.
Roaming
A moving device handing over from one access point to the next. It is decided by the device, not the network, which is why cells have to overlap by design — particularly for scanners, voice handsets and AGVs.
Predictive survey
A design modelled from your drawings and a wall-material schedule, before the building exists or before anything is bought. Accurate enough to budget and procure from.
On-site survey
Measurement taken in the building as built and as loaded, capturing what a model cannot know: substituted materials, stock, machinery, and the neighbours' networks.
AP-on-a-stick
A survey access point mounted temporarily on a tripod at a candidate position and height, so real propagation from that exact spot can be measured rather than predicted. The step that settles arguments about high-bay warehouses and awkward buildings.
Validation survey
A survey run after installation and commissioning, comparing the built network against the design. The only step that proves a coverage promise was delivered.
Heatmap
Measured radio data painted over a scaled plan. It means nothing until you know which measurement is plotted and at what threshold — the same data can look excellent or alarming depending on where the scale is set.
Bill of materials
The costed equipment list at the back of the report, with part numbers, so the work can be tendered on identical scope.
PoE
Power over Ethernet — the access point draws its power from the same cable that carries its data, which is why switch capacity and PoE budget are part of a Wi-Fi design.
SSID
The network name devices see. More SSIDs is not better: each one costs airtime on every radio broadcasting it.
Wi-Fi 6 / 6E / 7
Successive standards (802.11ax and 802.11be). 6E and 7 add the 6 GHz band, which carries far more capacity but travels less far and tolerates walls less well — so a 6 GHz design often needs more access points than the 5 GHz one it replaces, not fewer.
Common questions
Is a Wi-Fi survey the same as a network audit?
No. A network audit looks at switching, routing, security and configuration. A Wi-Fi survey looks at radio: coverage, noise, capacity and roaming. They overlap — capacity problems are sometimes an uplink problem wearing a Wi-Fi costume — and where we find that, the report says so plainly.
Can't we just use a phone app?
An app tells you what one radio, held at chest height, sees in one direction, at one moment. It cannot show you the noise floor, retry rates, channel utilisation, what your neighbours are transmitting, or whether a scanner will roam cleanly down an aisle. It is useful for a quick sanity check and useless as the basis for a design you are going to spend money on.
How is a survey different from a site visit?
A site visit produces an opinion. A survey produces measurements, a design and a document — one that any competent installer can build from and that you can hold somebody to afterwards.
Does the building have to be empty?
The opposite, usually. An empty warehouse measures the easy case; loaded racking attenuates far more. A capacity problem can only be measured with the people there. Where it matters we survey twice — once empty for the physics, once loaded for the behaviour.
How long does a survey stay valid?
Until the building or the load changes materially. The list above covers what shifts it. A site that is operationally dependent on Wi-Fi is worth re-checking annually; one that has just had a mezzanine put in should be re-checked straight away.
Who should commission one?
Whoever owns the consequence of it not working — usually IT, facilities or operations. You do not need to be technical to commission or read one. That is the point of the plain-English summary at the front.
Ready to scope one?
The pre-survey questionnaire takes about ten minutes and usually gets you a fixed price without a discovery call.