Buying Legrand Smart Lighting for an Office? What I Check Before I Sign the PO

If you're putting Legrand smart lighting into an office, three things decide whether the project works, and none of them are the switch plate finish: where the Zigbee gateway sits, what LED driver is spec'd behind the fixture, and who actually owns the app once the installer's van pulls away.

I've run this purchase twice. The first time I optimized for unit price and spent the next eight months cleaning up after it. The second time cost roughly 18% more on paper and we haven't had a callback yet. That gap is basically the whole article.

I'm the office administrator for a 380-person company — call it 400 if you count contractors — and I manage facilities purchasing across three sites. Around $1.1M a year, spread over maybe a dozen vendors. I report to operations, and finance reviews anything over $5,000. So when I say "cheapest quote," I mean the number that gets the PO approved, not the number that keeps the lights behaving in February.

March 2023 changed how I think about lighting spec sheets. We were re-lamping a 4,000 sq ft open floor and got three quotes. The middle one was about $1,900 cheaper on the driver line item. I took it. Nine months later, 11 of the 40 fixtures were flickering or dead. The driver order was $5,800 — actually $6,200 with freight, I keep forgetting the freight. Replacement labor and disposal came to around $8,900, give or take. Plus two months of "why is the conference room strobing" tickets landing in my inbox.

The door sensor is the cheapest part and the easiest to get wrong

A Legrand smart door and window sensor in an office is not a security device. Treat it as a scene trigger. Ours sit on three conference rooms, the print room, the server closet, and two storage cabinets nobody wants to walk to.

Two things bit us. First, placement. The contact gap tolerance is tighter than people assume, and a door that swings a little on its hinge will report open when it's closed. We had a conference room that logged 30 "door opened" events a day until I moved the magnet 4mm. Second, and this is the one that surprised me: battery-powered Zigbee sensors are almost always sleepy end devices. They don't route traffic for anything else. So scattering sensors down a hallway does nothing for your mesh. Never expected that. Turns out all those little sensors I was proud of were doing zero network work.

If a rep tells you "more sensors means a stronger network," that's a red flag, not a feature. Routers need mains power. Plan the gateway placement around the powered devices — switches, dimmers, plugs — and let the sensors be passengers.

Somebody has to own the Legrand lighting control app

Here's the part nobody puts in the quote. The Legrand lighting control app needs an owner inside your building, not just an installer login. We found out the hard way: a contractor's account still had admin rights three months after handover. Nobody had revoked it because nobody had written down that it existed.

What I ask now, before signing:

• Who is the named administrator, and what happens when they leave?
• Does it support SSO or at least per-user accounts, or is it one shared password on a sticky note?
• What still works if the app is down or the internet is out?
• Is there a documented API, and does IT get read access to it?

The surprise on round two wasn't the app's interface. It was how much IT cared about the API and local fallback versus how little they cared about the UI. Our network lead spent ten minutes on the screens and forty on the integration docs. Wall switches still work if the gateway dies — that was the deal-breaker question for him, and honestly it should have been mine too.

Zigbee device certification covers less than you think

Every Zigbee device in your project should carry Zigbee 3.0 certification. That's table stakes. But certification tests a defined feature set, not every optional attribute a vendor decides to expose. Two certified devices can still surprise you on dimming curves, reporting intervals, or how they behave after a power cut.

Per the Connectivity Standards Alliance (csa-iot.org), Zigbee 3.0 certification unifies device types under a single application layer and standardizes commissioning. It does not guarantee that two certified products implement every optional cluster attribute identically.

Practical translation: build a compatibility list before you buy, not after. Ours is one page. Device model, firmware version, gateway firmware, who tested it, and the date. Boring, and it has saved me twice.

On the zigbee api side — if you're tying lighting into a BMS or a room-booking system, ask for the actual API documentation during the quoting phase. Not a datasheet that says "API available." The real document, with the endpoints and rate limits in it. Two vendors I talked to couldn't produce it until after the PO, which told me everything.

"What LED driver do I need" is the question that decides your budget

This is where the money actually goes, and it's the line item everyone skims. The driver isn't glamorous, so it becomes the place people shave $40 a fixture and then pay $220 a fixture to fix it.

The sequence I use now:

1. Start with the dimming protocol, not the fixture. Phase-cut (ELV/TRIAC), 0-10V, DALI, or Zigbee-native control each need a different driver. Buying the fixture first and matching a driver later is how you end up with a dimmer that only goes down to 15%.
2. Constant current or constant voltage. Get this from the fixture datasheet, not from a chat with a sales rep.
3. Leave headroom. I spec drivers at roughly 80% of rated load. Running at 100% in a ceiling plenum in August is asking for early failure.
4. Check the listed environment. Ambient temperature, derating curve, and whether the driver is rated for the enclosure you're putting it in.
5. Look for UL 8750 listing and Class 2 output where the wiring method requires it. Under NFPA 70 (NEC), Class 2 circuits have specific installation rules, and a driver that isn't listed for your application is a code conversation you don't want.
6. Test the low end. Ask for the minimum dim level in writing, then go stand in the room and look at it. Numbers on a sheet and what your eyes see at 5% are not the same thing.

Our failure was item six. The cheap driver claimed compatibility with "most 0-10V dimmers." It was compatible. It just bottomed out at 15% and buzzed. For a room used for video calls, that's not a minor issue.

How I run the math now

Bottom line: I stopped comparing unit prices and started comparing what I call the six-line total.

• Fixture and driver, landed cost
• Labor and commissioning
• Spares — I now buy 5% spares up front, which nobody likes in the budget meeting
• Rework risk, priced at what it cost us last time
• Who fixes it at 2am, and at what rate
• Administration — app seats, firmware updates, someone's time

On energy claims specifically: ask for the underlying data. Under the FTC's Green Guides (16 CFR Part 260, ftc.gov/green-guides), environmental claims including energy savings have to be substantiated with competent and reliable evidence. So when a vendor slides a number across the table, the fair question is "what's the methodology and the baseline?" Nobody has ever been offended by that question. A few have been unable to answer it, which is also an answer.

That said — lighting controls do save energy in the right rooms. Occupancy sensing in low-traffic areas, daylight dimming near windows, scheduling for after-hours. Whether your specific floor saves anything meaningful depends on how people actually use it, and I'd rather see your utility data than a brochure percentage.

When the cheaper route is genuinely fine

I don't want to overcorrect here. There are rooms where the inexpensive driver and the basic gateway are the right call. A single storage room, a stairwell, a back-of-house corridor nobody dims. If there's no BMS integration, no scene programming, and a local electrician on retainer who can swap a driver in a day, the premium doesn't buy you much.

And I should be honest about my limits: I'm a purchasing person, not an electrical engineer. Dimming curves, flicker percent, and harmonic limits are conversations I have with our consultant, not calls I make alone. What I do own is the question list — and the part where I decide whether the cheapest number on the page is actually the cheapest project. In my experience across three sites and two rounds of this, it usually isn't.