RF Site Survey: A Standard Operating Procedure

Purpose

This document is deliberately ungated and written to be adapted. Replace the thresholds with your own standards and issue it to your team or your contractor.

Survey types

Predictive. Modelled from floor plans. Only as good as its material assumptions. Always the starting point, never the only step.

Validation (AP-on-a-stick). A single AP placed at candidate locations, measured live. Strongly recommended for any site with unusual construction.

Post-deployment. Measured on the installed, configured network. The only survey that proves anything. Mandatory before sign-off.

Passive. Listening only — RSSI, noise, SNR, channel utilisation, AP visibility.

Active. Associated to the network — throughput, retry rate, roaming behaviour.

Equipment

Survey software with floor plan import and 6 GHz support.
A representative client device — survey adapters have better antennas than real laptops and flatter results.
Spectrum analyser for non-Wi-Fi energy.
Measuring wheel or laser measure for scale verification.
Camera. Photograph every AP location and every unusual construction detail.
Tripod, pole and battery for AP-on-a-stick.

Method — post-deployment passive survey

Verify the floor plan scale against a physical measurement.
Confirm the network is in its final configuration — channel plan, power, widths, all SSIDs enabled.
Walk at a steady pace, roughly 1 m/s, pausing 5–10 seconds at each marked point.
Cover the perimeter first, then a serpentine pattern with passes no more than 5 m apart.
Include stairwells, lifts, plant rooms and external areas, or explicitly document them as out of scope.
Survey each band separately if the tool cannot capture simultaneously. Never extrapolate 6 GHz from 5 GHz.
Note occupancy. An empty-room survey of a space that will hold 200 people is optimistic by several decibels. The same applies to warehouse racking — survey it loaded.

Measurement thresholds

Metric	Primary coverage	Secondary	Fail
RSSI (design band)	≥ -67 dBm	-67 to -72 dBm	< -75 dBm
SNR	≥ 25 dB	20–25 dB	< 20 dB
Channel utilisation	< 40%	40–60%	> 70%
Retry rate	< 10%	10–15%	> 20%
APs audible at ≥ -75 dBm	2–3	4	5+ (excess overlap)
Cell overlap at threshold	15–20%	20–30%	> 30%

Method — active survey

Associate to the production SSID with production credentials.
Run iperf3 or equivalent to a wired server, upstream and downstream, at a minimum of six representative points including two at the coverage edge.
Walk a defined roaming route with a continuous ping and a live call. Record every roam, the RSSI at which it occurred, and any audible interruption.
Test with the worst client class you support, not the best.

Errors that invalidate a survey

Wrong floor plan scale.
Surveying with an adapter that outperforms every real client on site.
Surveying an empty space that will be occupied, or unfurnished space that will be furnished.
Surveying a warehouse with empty racking.
Extrapolating one band's results to another.
Walking too fast, producing sparse data interpolated into a confident-looking heat map.
Surveying before the channel plan and power levels are final.
Omitting a spectrum capture, and therefore missing non-Wi-Fi interference entirely.

Deliverables

Heat maps per band: RSSI, SNR, channel utilisation, AP coverage, predicted data rate.
Spectrum capture summary with interferers identified and located.
Throughput results table with test locations marked.
Roaming test log.
Photographs of every AP location and notable construction feature.
Exceptions list: areas failing threshold, with a recommendation and indicative cost for each.
A one-page summary a non-engineer can act on.

Sign-off criteria

The survey passes when every in-scope area meets primary or secondary coverage thresholds at the design band, no area exceeds the channel utilisation fail threshold at peak, the roaming route completes with no audible interruption, and every exception is documented with an agreed remediation or an accepted risk.