-
How a 34-room hotel project started, and how I assumed it would go
-
The first sign of trouble: crystal, dimmers, and a fluorescent-era assumption
- Where the cordless chandeliers nearly sank the opening night
-
The size that didn't fit, and the batch that didn't match
-
What it actually cost, and what changed
-
Where I'd do it differently — and where I wouldn't
How a 34-room hotel project started, and how I assumed it would go
February 14, 2023, 4:40 p.m. I was sitting in my car in a hotel parking lot in Richmond with a 96-page lighting spec book open on the passenger seat. The FF&E designer had sent her final selections that morning, and I had one week to turn them into purchase orders.
I've been handling commercial lighting orders for 11 years. In that time I've personally made — and written down — 17 significant mistakes, roughly $23,000 in wasted budget across those projects. This is the story of mistake #14. It cost us $9,400. It's also the reason our team now runs a 31-line pre-order check on every single fixture schedule, no exceptions.
The project was a 34-key boutique hotel renovation. Small enough that the owner knew every line item. Big enough that a $9,400 error showed up on a board report.
The lighting scope broke into four buckets:
- Lobby and suites: designer crystal chandeliers — visual comfort chandelier crystal pieces the designer had used on two previous projects and trusted.
- Restaurant: a visual comfort bonnington chandelier over the 12-seat communal table. It was the one fixture the owner personally signed off on.
- Bar lounge: an eichholtz chandelier-inspired statement piece with a soft, sculptural silhouette.
- Rooftop pergola and corridor: cordless chandelier fixtures for the pergola (no conduit, no junction boxes, historic structure) and LED panel lights for corridors and back-of-house.
On paper it was a clean schedule. Four categories, clear quantities, one supplier relationship. I remember thinking this would be the easy project of the quarter. I was fairly wrong.
The first sign of trouble: crystal, dimmers, and a fluorescent-era assumption
The crystal chandeliers went in first. Twelve of them. They looked right when they arrived — the packing was excellent, nothing broken, and the crystal had that weight to it that you can feel through a cardboard box.
Then the electrician called. Three of the fixtures were buzzing at low dim levels, and the two in the suites flickered below 20%.
Here's the part that still embarrasses me. My assumption was that dimming is dimming. That came from an era when most commercial fixtures ran on magnetic low-voltage transformers and a basic forward-phase dimmer handled pretty much anything. That's changed. Modern LED-integrated and low-voltage fixtures have drivers with specific dimming protocols — forward phase, reverse phase, 0-10V, or a proprietary system. The fixture doesn't care what the box on the wall says. It cares whether the driver and the dimmer speak the same language.
Our spec book named the fixture. It didn't name the driver compatibility. So the electrician installed the dimmers he had on the truck, and the fixtures did exactly what they were designed to do when fed the wrong control signal.
The fix: two new dimmer banks, one re-pull of control wiring in the suite corridor, and $1,850 in electrician time that wasn't in anyone's budget. And I still had three buckets of the schedule to place.
Where the cordless chandeliers nearly sank the opening night
The rooftop pergola was the fun part of the project — a 1950s structure with beautiful exposed brick and absolutely no way to run conduit without tearing into it. The designer proposed cordless chandelier fixtures. Battery-powered, rechargeable, no wiring. On a historic structure with a hard budget line for structural work, it was the right call.
What I did wrong was this: I read the runtime spec, saw "up to 10 hours," and mentally checked the box. Done.
Nobody asked the two questions that actually mattered. First, at what brightness? Runtime drops as you push output up, and "up to 10 hours" usually describes the lowest usable setting, not the setting a designer wants for a rooftop bar at dusk. Second, what happens at hour seven? Lumen output tapers as the battery drains. A fixture that looks fantastic at 8 p.m. can be noticeably dimmer by midnight, and if your event runs to 1 a.m., the last hour is the one people remember.
We tested four fixtures on site in March, ran them from 7 p.m. to 1 a.m., and measured. At the output level the designer wanted, we were getting closer to five and a half hours on a full charge. For a wedding that runs long, that's a problem.
So we did what I should have done at the quote stage: we bought double the fixture count and staggered the charging cycle so half the units sat on standby in a back-of-house cabinet, fully charged, ready to swap in a 90-second changeover during dinner service.
That decision cost $2,900 in rush freight on the additional units. But it saved the opening weekend.
The part where my gut beat the spreadsheet
We also had to source the LED panel lights — corridors, back-of-house, employee areas. That's about 90 panels, which is not glamorous but it's 90 line items.
I ran the numbers three ways. A marketplace seller was quoting roughly 40% less than the local electrical distributor we'd used for years. Same wattage, same form factor, same claimed specs on the data sheet. Every spreadsheet analysis pointed to the marketplace option: it saved about $1,900.
Something felt off. Their product page listed a DLC model number, but when I emailed to ask for the specific listing certificate, the reply took four days and the document didn't match the SKU.
I split the order. Half from the distributor at full price, half from the marketplace seller, because I didn't want to explain to the owner why I'd paid $1,900 more "on a feeling."
Turns out the feeling was right. The marketplace panels measured around 78 CRI and flickered visibly at 10% on our dimming test. The distributor panels were clean. We ended up returning the cheap half and buying the rest from the distributor anyway — plus a restocking fee.
That's not a story about cheap products being bad. Most budget panels are fine for a stockroom. It's a story about cheap panels not being fine next to a $4,000 crystal chandelier in a guest corridor where the color temperature shift is immediately visible.
"Where can I buy an LED panel light?" is the easiest question in this entire project, and somehow it's the one that cost us the most.
The complication is what happens after you decide where to buy. For a commercial project, the panel needs a proper listing (UL 1598 or an equivalent ETL/CSA mark is the baseline for luminaires in the US). Photometric performance should be backed by an IES LM-79 test report, and if you're going after utility rebates, the model needs to appear in the DesignLights Consortium qualified products list. At the time, DLC 5.1 was the active technical requirement version, and the local utility rebate program keyed directly off it.
Our first panel submittal wasn't on the qualified list. We lost a $1,200 rebate. Not because the panel was bad — because I never asked for the listing certificate before the purchase order went out.
The size that didn't fit, and the batch that didn't match
Two smaller issues, both preventable, both my fault.
The eichholtz chandelier-spec piece for the bar lounge measured 47 inches at its widest point. Our ceiling drop allowed 44 inches before it would clip the sightline from the mezzanine. I'd checked the diameter against the plan view but not against the elevation. The replacement rod arrangement cost us a week and some awkward emails.
And then there was the finish. Hand-applied plaster and alabaster are natural materials. Two batches, even from the same collection, will not match perfectly under raking light. We ordered in two waves because of the budget cycle, and when the second wave arrived the lobby read as two slightly different rooms. To be fair, the client noticed before I did — she stood in the doorway and asked why the right side looked "warmer."
The honest answer is that this is normal for natural materials, and the fix is ordering the full quantity in a single production run and specifying that in writing. That's what we do now. It's line 19 on the checklist.
What it actually cost, and what changed
Here's the tally, since I made the team write it down:
- $1,850 — electrician rework on dimmer compatibility
- $2,900 — rush freight on additional cordless chandelier units
- $1,200 — forfeited utility rebate
- $2,300 — restocking fees and return freight on the panel and rod issues
- $1,150 — roughly 19 hours of PM and coordinator time on reorders
Total: about $9,400, plus a six-day schedule slip and one very uncomfortable conversation with a designer who had trusted me.
The checklist that came out of it has 31 lines. The ones that would have caught this project:
- Dimmer/driver compatibility confirmed in writing — protocol, load, minimum load — before the fixture PO.
- Battery runtime verified at the intended output level, with the event schedule on the table, not just the data sheet.
- DLC or JA8 listing certificate received and filed before purchase. Particularly on projects with rebate exposure.
- Overview dimensions checked against both plan view and elevation, with the drop dimension measured on site.
- Natural-material fixtures ordered in one production run, with a written note that dye lot and material variation are expected.
Where I'd do it differently — and where I wouldn't
I want to be honest about the limits here, because the recommendation cuts both ways.
If you're specifying a lobby, a restaurant, or any space where people will actually look at the light, going with a designer collection — a visual comfort chandelier crystal piece, the bonnington, a sculptural eichholtz chandelier — is usually worth the premium. Not because the crystal is objectively better, but because the client bought a design, and the fixture is the design. On that project, the bonnington chandelier over the communal table is the single thing guests photograph most. That part worked.
But I'd push back on that recommendation in three specific cases. If the space is back-of-house, a stockroom, or a mechanical corridor, buy the DLC-listed panel from your distributor and stop overthinking it — nobody is going to study the beam angle at 2 a.m. during a linen delivery. If the schedule is inside four weeks, made-to-order designer fixtures are the wrong tool regardless of how good they look. And if the cordless chandelier installation is going to run more than five hours a night, every night, on a fixed battery cycle, I'd go back to the designer and ask for a hardwired option first, even if it costs structural work.
Even after we locked the cordless decision, I kept second-guessing it. What if the changeover during dinner service looked like chaos to a guest? The four weeks between committing to the double order and opening night were stressful. I didn't relax until the first Saturday event ran to 12:50 a.m., the swap happened in under two minutes, and not one person in the room noticed.
That's the thing about these mistakes. None of them were about picking a bad product. Every single one was a missing question — dimming protocol, runtime at brightness, listing certificate, elevation dimension, batch consistency. The fixtures were fine. My process wasn't.
The checklist is longer now. So is the time I spend on the front end of an order. I'll take that trade.

