September 2024, 6:47 AM
I was on my second coffee, scrolling through the seventh lighting quote in our procurement system, and I did the math three times because I didn't believe it.
Lowest bid: $68,400. Second-lowest: $87,300. Spread: nearly $19,000 on a 420-device lighting retrofit for our four-story office.
My director had already forwarded the cheapest one with a note: "Looks like a winner?"
I almost agreed. Almost.
How this started
I'm a procurement manager at a 150-person insurance services firm. I've owned our facilities and capital expense budget — about $1.1M annually — for six years. That means every switch, every sensor, every recessed fixture that goes into this building has come across my desk at some point.
Last fall, we finally got board approval to replace our 22-year-old fluorescent system with LED plus occupancy sensing and dimming. Budget: $98,000. Scope: roughly 420 points across offices, corridors, and two conference floors.
I thought this would be straightforward. Buy switches, buy sensors, buy panels, get an electrician.
It was not straightforward.
The trap in the $68,400 quote
I spent two weeks on supplier screening. Seven made the short list — three were regional distributors, two were direct OEM reps, and two were online-first sellers who quoted suspiciously fast.
The cheapest bid broke down cleanly on paper:
- Wall switches: $2.13/unit × 230
- Occupancy sensors: $18/unit × 90
- Dimmers: $22/unit × 65
- Recessed fixtures: $31/unit × 35
Roughly $62K in product plus $6,400 "installation support."
Here's the thing. Most buyers focus on unit price and completely miss the line-item that isn't on the quote at all. I've been burned by that before — a $3,800 "free setup" offer in 2022 that ended up costing us $1,240 in rework fees. So this time, before I signed anything, I asked for spec sheets on two specific items: the bulk zigbee module they were proposing, and the recessed lighting OEM they were sourcing from.
The reply took four days. When it came back, one line stopped me cold:
"Firmware is compatible with major lighting control protocols."
That's not a spec. That's marketing.
The question I should have asked first
The question everyone asks a supplier is "what's your best price?"
The question they should ask is "what's included in that price, and what changes when 420 devices become 420 devices that all need to talk to each other?"
Because here's what I learned. A switch isn't a switch. A sensor isn't a sensor. The moment you put more than about 50 zigbee nodes on a single gateway, things get interesting — and not in a good way.
Our building needed full coverage. That's 420 nodes. I called three of the seven suppliers back and asked one specific question: how many nodes per gateway, and does the gateway cost come out of your quote or mine?
Two didn't know. One hedged with "typically it depends."
Only three gave me a straight answer with a documented number.
The Legrand occupancy sensor wiring conversation
I'll be honest about where I stumbled. When I got into the technical weeds on occupancy sensor wiring — specifically the legrand occupancy sensor wiring setup one of our electricians recommended we reference — I realized I'd been thinking about the wrong problem.
I'd been focused on device compatibility. What I should have been focused on was wiring topology and control mode default.
Here's what that looked like in practice: our corridors and stairwells needed manual-on/auto-off behavior. Most factory defaults ship as auto-on/auto-off. That's a configuration issue, not a product issue — which means it doesn't show up on the quote, but it absolutely shows up on the invoice when the electrician has to come back.
Two of the seven quotes included this in writing. Five did not.
The real TCO conversation
I built a spreadsheet. I'm not a data analyst — I'm just someone who's been burned enough times to track every line.
When I loaded all seven quotes into the same format — product, installation, required infrastructure, commissioning time, and a 24-month maintenance buffer — the spread collapsed from $19,000 to about $7,800.
Still significant. But not the same decision.
The $68,400 quote became $81,760 once I added:
- Two additional gateways ($960)
- Rework of five sensor zones ($4,700 in electrician time)
- Firmware commissioning from vendor technicians ($2,200)
- Replacement of 14 switches that failed a compatibility check with our building management system ($5,500)
Your math may land somewhere different than mine. If you're working with under 100 devices, my sense is the gateway math barely matters. If you're past 400 — which is where most mid-size office retrofits land — it becomes the whole conversation.
What actually decided it
We didn't go with the cheapest. We didn't go with the most expensive either.
We chose a mid-range supplier — total landed $92,000 — because they were the only one who sent us a physical sample unit before signing. Not a data sheet. An actual switch and an actual sensor, configured the way we needed, with a one-page commissioning note.
That's the thing about what to look for in a ceiling light supplier that almost nobody says out loud: the sample is the tell. If they can't or won't ship a working configured unit before you commit to a $90K PO, everything after that is a guess.
Results, six months in
Installation ran three weeks. Two hiccups — one batch of fixtures arrived with a firmware version that didn't match what was on the spec sheet, and we had to re-flash 35 drivers. Took two days. Not free, but manageable.
Energy draw dropped 27% against the same quarter last year. That's a real number from our utility portal, not a projection.
Failure rate so far: 2 switches out of 420. Both replaced under warranty, both within 30 days.
Three things I'd tell you if you're doing this next quarter
One. Get a physical sample. Every time. Vendors who can't ship one are telling you something.
Two. Ask for node counts, gateway counts, and firmware version numbers in writing before you scope. "Compatible with major protocols" is not a specification — it's a hope.
Three. When a supplier references Legrand-style specs as their quality benchmark, ask them what they actually tested against. Benchmarks are fine. Evidence of testing is better.
I don't have hard data on how any of this scales past 500 nodes or into multi-floor industrial settings. My experience is 420 devices, one building, one quarter of real usage. That's the sample size. Treat it that way.
But the $19,000 gap between quote 1 and quote 2? That gap was never real. It was just the cost of what wasn't on the page yet.