Wireless Site Surveys: Designing Wi-Fi That Works the First Time

Wi-Fi problems are almost always design problems

When a wireless network is slow, drops calls, or won't hold a roam across a floor, the instinct is to blame the access points or add more of them. More often the fault is upstream of the hardware: the network was designed by eyeballing a floor plan, spacing access points on a convenient grid, and turning them to full power. Radio does not respect convenience. A single glass wall, a metal shelving run, a microwave oven, or a neighboring tenant's network can undo a deployment that looked fine on paper — and adding more radios to a badly designed cell usually makes it worse, not better, by raising the noise floor everyone shares.

A site survey is the discipline of replacing guesses with measurements. It answers, for a specific building with its specific construction and its specific users, where radios go, what channels and power they run, and whether the result actually meets the requirement — before the users arrive to discover it doesn't.

Coverage is easy; capacity is the real target

The oldest mistake in wireless design is treating it as a coverage problem — draw circles until the whole floor is green. That worked when a site had a laptop or two per room. It fails now because the constraint is capacity, and capacity is governed by how airtime is shared, not by how many bars a client shows.

Wi-Fi is half-duplex and every device on a channel — yours and your neighbor's — contends for the same airtime through carrier sense. A single access point cranked to maximum power blankets a wide area with a strong signal, but it forms one large collision domain in which hundreds of devices take turns through one radio. The capacity-oriented design does the opposite: smaller cells at moderate power, more access points each serving fewer clients, so airtime is divided among many radios running on non-overlapping channels. Design to the density and application profile — a lecture hall, a warehouse, and an open office with constant video calls impose completely different airtime demands on the same square footage. Coverage without capacity produces full signal bars and unusable throughput.

The four kinds of survey, and what each one sees

"Site survey" is really a family of techniques, used at different stages, that see different things. A serious design uses several.

• Predictive survey. Done in software before anything is installed. A calibrated RF model of the building — walls, materials, attenuation values — predicts coverage, channel plan, and access-point count from a floor plan. It is the design starting point, only as good as the material assumptions fed into it, and never a substitute for measuring the real building.
• Passive survey. A technician walks the space with a scanner that listens to all networks without associating, mapping actual signal strength, co-channel and adjacent-channel interference, and coverage as it truly is. It is the honest picture of the RF environment, including everyone else's networks.
• Active survey. The survey client associates to the network and measures what a real user experiences — throughput, retransmissions, roaming behavior, and round-trip latency — rather than just signal presence. This is how you catch a cell edge where the signal looks adequate but the data rate has collapsed.
• Spectrum analysis. A dedicated spectrum view sees non-Wi-Fi energy that a Wi-Fi scanner cannot: microwave ovens, cordless phones, video senders, Bluetooth, motion sensors, and radar. On 5 GHz, radar detection matters because DFS channels force an access point to vacate on detecting radar, and unexplained disconnects often trace back to exactly that.

Skipping the passive and active steps is how a predictive-only design ships with confident-looking heat maps and real dead zones.

Read the RF, not just the map

Good wireless design is fluent in a handful of physical realities that a floor plan hides:

• Materials attenuate differently. Drywall costs a few decibels; concrete, brick, metal, mirrored glass, and tile with foil backing cost far more. Two rooms the same distance from a radio can differ by 20 dB because of what is in the wall.
• Signal strength is not the whole story. The signal-to-noise ratio governs the data rate a client will actually negotiate. A strong signal in a noisy band performs worse than a moderate signal in a clean one, which is why raising power to "fix" coverage frequently backfires — it raises noise for every neighboring cell.
• The 2.4 GHz band is crowded and narrow. It offers only three non-overlapping 20 MHz channels (1, 6, 11) and is full of legacy and non-Wi-Fi interference. The 5 GHz and 6 GHz bands provide many more channels and cleaner air — the reason modern designs push clients onto them and use 2.4 GHz as a fallback.
• Channel width is a tradeoff. Wide channels (40/80/160 MHz) raise peak throughput but consume scarce spectrum and increase the odds of overlap. In dense deployments, narrower channels and more reuse usually beat a few very wide ones.

Validate, then keep watching

A survey is a hypothesis until it is proven in the installed environment. After the access points are mounted and cabled — on structured cabling that can also carry their power over PoE — run a post-installation validation survey with the real gear at production settings and confirm the design intent against the actual result:

• Coverage and SNR meet the target in every required area, including edges and stairwells.
• Cell overlap is enough for seamless roaming (clients hand off before the old signal dies) but not so much that co-channel interference spikes.
• Real throughput and latency meet the application requirement under representative load, not on an empty floor.
• The channel and power plan holds when neighboring networks and DFS events change the environment.

The RF environment is not static — new tenants, new walls, new devices, and firmware changes all shift it — so wireless needs ongoing infrastructure monitoring of channel utilization, retries, roaming failures, and interference, not a one-time sign-off.

Where to start

Begin with the requirement, not the hardware. Document the client density, applications, and roaming needs per area; those numbers, not a coverage radius, drive the design. From there:

1. Build a predictive model from an accurate floor plan with real material values.
2. Validate it with passive, active, and spectrum surveys in the actual space.
3. Design for capacity — smaller cells, moderate power, a clean channel plan favoring 5/6 GHz.
4. Run a post-install validation survey and fix the gaps before go-live.
5. Monitor the RF environment continuously and re-survey after major changes.

Wireless that works the first time is not luck; it is measurement applied before installation instead of complaints applied after. intSignal surveys, designs, and validates enterprise Wi-Fi so coverage and capacity both hold under real use — talk to our team to design a wireless network you won't have to apologize for.