Site surveys: the things you measure that aren't on the spec sheet
The fastest way to lose money on an IoT deployment is to skip the site survey. The second fastest is to do one and only measure the things on the spec sheet.
A spec sheet tells you what the hardware can do under lab conditions. A site survey tells you whether it will do that here, mounted there, talking to those things, in this weather, with that ground.
Here’s what we actually measure when we send a tech to a site before quoting an install.
The mounting location, not the site
Customers will walk you to the office trailer, the gate, and the breakroom. None of those are where the gateway goes. The mounting location is on the roof, on a pole, on the side of the equipment shed — typically the highest spot with line-of-sight to whatever the gateway needs to talk to.
We don’t take signal readings until we’re standing where the antenna will be. The difference between a reading at the front door and a reading on the roof can be 15 dB. That’s the difference between “this works” and “this doesn’t.”
Cellular: every carrier, three numbers
We carry a multi-carrier signal meter. At the mounting location, for every carrier we have a SIM for, we record:
- RSRP — raw signal strength. Anything weaker than -110 dBm and we plan for an external antenna at minimum.
- RSRQ — quality. Tells you whether the signal is clean or fighting interference from the same tower.
- SINR — signal-to-noise. The number that actually predicts throughput. Below 0 dB and you’re going to see retransmissions even with full bars.
We don’t decide based on “bars.” Bars are a UI element, not a measurement.
RF coverage: walk the footprint
For BLE and LoRaWAN deployments, the gateway location is half the answer. The other half is whether the things you need to track are actually inside the gateway’s coverage envelope.
We walk the footprint with a known-good test beacon and a handheld scanner, recording RSSI at the corners, in the dead spots behind metal racking, and at any point a customer specifically wants tracked. Datasheet range numbers don’t survive a metal yard, a stack of pallets, or a rebar-heavy concrete wall.
If the customer’s actual workflow happens in a coverage hole, we find out now, not after the install.
Power: the breaker, the load, the ground
The wall outlet is the easy part. What we want to know:
- Which breaker is this on, and what else is on it? A gateway sharing a circuit with a welder is a gateway that reboots every time someone strikes an arc.
- What’s the available load budget? Cellular routers, PoE switches, cameras, and the gateway add up faster than people expect.
- Is the ground actually grounded? A surprising number of “outdoor” outlets at industrial sites have a ground pin connected to nothing. We test it. If it’s not bonded, we plan for a separate ground rod.
Lightning is the single most expensive event on a deployment, and it’s the one customers most consistently underestimate.
Mount points: structural, not cosmetic
A wall is not a mount point until you know what’s behind it. We carry a stud finder and a long drill bit. The question we’re answering is: can this thing survive a 60 mph wind with an antenna on a mast attached to it?
If the answer is uncertain, we plan for an unistrut frame anchored to something structural. “We’ll just screw it into the side of the building” is a return trip after the first storm.
Network: who owns the path
For deployments that bridge to a customer’s network — uploading to their data lake, integrating with their SCADA, anything that isn’t pure cellular backhaul — we want to know:
- Who’s the IT contact, and are they aware this is happening?
- What’s the firewall posture? Can we reach our broker on 8883 outbound?
- Is there a NAT or proxy in the path that’s going to do unhelpful things to MQTT keepalives?
- Who has the credentials to add a device to their network if we need to?
Half the install delays we’ve ever had were “the IT person who knows the firewall is on vacation.”
Weather and seasons
The site we’re surveying in May is not the site that exists in February. We ask:
- Where does snow accumulate? Does it bury the antenna or load up the mounting bracket?
- Where does water pool? Is the proposed enclosure location below a roof drip line?
- What’s the seasonal foliage situation? A clear line of sight in March is a wall of leaves in July.
- What’s the temperature range? Lithium batteries don’t charge below freezing, and a sealed enclosure in direct sun in Phoenix can hit 70°C internally.
We’ve redone deployments because nobody asked about snow. Once.
People: the on-site contact
The last measurement isn’t technical. It’s: who is going to be on this site five days a week, and what’s their phone number?
When a gateway goes offline at 2am, we want to be able to call somebody who can walk over and look at it. If the answer is “the site is unmanned,” we plan for full remote diagnostics, dual-SIM cellular, battery backup, and a relay we can power-cycle the gateway with. If there’s a person on site, the install gets simpler.
Ask the question early. The answer changes the bill of materials.
What goes home with the tech
Every survey produces:
- GPS-tagged photos in every direction from the mounting location
- A signal report per carrier with RSRP/RSRQ/SINR
- An RF coverage map of the footprint
- A power and ground assessment
- A list of named contacts: site, IT, electrical
- Notes on weather risk and seasonal change
The deliverable is a one-page sheet. The work is two hours. The cost of skipping it is a return trip with the wrong hardware and a customer who’s lost confidence.
If you’re planning a deployment and want a sanity-check before the hardware ships, we’re happy to do the survey. It’s almost always cheaper than discovering the gap after the brackets are mounted.