It was a rainy Tuesday in late September when our facilities tech dropped a stack of photos on my desk. The parking lot at our manufacturing plant was becoming a liability. Three of the six metal halide fixtures were completely dead, the other three were humming along at maybe half output, and the night-shift crew had started asking uncomfortable questions about safety. Kinda hard to ignore.
As the procurement manager for a 40-person aerospace parts company, I manage about $80,000 a year in facilities spending. I've tracked every invoice for six years. And in that time I'd come to believe lighting was a commodity: compare the spec sheets, take the reasonable price, move on.
That was the mindset I carried into the parking lot upgrade. It turned out to be exactly the wrong mindset. It cost me nearly $10,000 to learn that.
The Cheap Quote Looked Great on Paper
I sent RFQs to seven vendors in early October. Six responded within a week, and the price spread was wider than I'd ever seen on a comparable product. A Cree quote came in at $14,200 for six units. A no-name importer out of Texas quoted $7,900 — the same lumen count, same IP65 rating, same five-year warranty language. The other four sat somewhere in between.
I did my due diligence, or so I told myself. Side-by-side spec spreadsheets. Three reference calls for the importer, all positive. Sign-off from finance. We placed the order on October 16th. Fixtures arrived November 8th, and the electrical contractor had them mounted in two days. That Friday evening I stood in the lot while all six units lit up the asphalt, and I took a picture to send to the CEO. I was pretty proud of myself.
Fast forward to January 17th. 9:40 PM. A text from the night supervisor: two of the six fixtures were flickering. By noon the next day, one was stone dead.
Okay, I thought. Warranty. But the warranty attached to that cheap quote was not the warranty I'd imagined. To file a claim, I had to ship the failed unit back to Texas at my own cost — $86 in freight for a 40-pound fixture — and their shop would evaluate it. Quoted lead time: four to six weeks. Meanwhile, a corner of the lot sat dark, and two employees reported nearly backing into each other.
(Ugh. The word "warranty" was doing a lot of heavy lifting in that purchase order.)
Over the next two months, three more units started acting up. Not a clean failure — more of a flicker-for-an-hour, quit-for-a-day, suddenly-work-again thing. My contractor came out twice and confirmed the wiring and mounting were fine on our end. I emailed the importer nine times and got useful responses to three. Each one asked again for the serial numbers and the purchase date. We had already sent them.
This is the part I want you to actually hear. The cheap fixtures cost $7,900. The repair calls cost $1,100. Freight for warranty returns cost $340. Disposal of the dead units cost another $270. That's $9,610 spent on a product that still wasn't doing the job. If I'd taken the Cree quote in October, the fixtures plus labor would have totaled around $16,100. Instead, I was looking at a path that ends at $25,710. I saved $6,300 on the invoice and watched it disappear on the back end.
The Turning Point: Data, a Cree Demo, and a New Office Question
One evening in March, I sat down with my Mac, opened Spotlight, and pulled up every vendor email and invoice from the past three years of facility projects. The pattern was embarrassing. Every project where we'd taken the lowest quote had generated a follow-up problem within six months. Every project where we'd paid for a mid-tier or premium product had been boring and trouble-free. I do not mean "slightly better." I mean zero follow-ups versus six. The spreadsheet didn't care about my budget-saving reputation.
Remember that $14,200 Cree quote from October? It was still in my inbox. I'd dismissed it as brand-name tax. So I called the distributor who'd sent it and asked if he'd be willing to do something a little unusual. He brought a sample fixture to our shop (this was back in March 2025), and we ran it on the bench for 72 hours straight. I measured the surface temperature against the import unit we'd already installed. The Cree fixture ran about 30°F cooler. That's not cosmetic. Heat is what kills LED drivers, and a cooler-running driver is a longer-running driver.
The spec sheet listed a projected L70 of 100,000 hours — the expected point at which a fixture still produces at least 70% of its initial brightness. And the warranty was structured differently from the import's: an advance replacement unit sent to us before we had to return the failed one. That single term told me more about vendor confidence than any marketing page could. We ordered six Cree LED parking lot light fixtures in early April.
Around the same time, we started planning the lighting refresh for our 8,000-square-foot office and engineering area. The old 2x4 troffers were past their useful life, and the new layout called for recessed lights. That's when I hit a question I hadn't expected: canned vs canless recessed lighting.
If you haven't dealt with this one, here's the short version. Canned lights use a recessed housing installed in the ceiling, and a trim plus a bulb or retrofit module sit inside that housing. Canless fixtures — often called wafer lights — are slim, self-contained units that mount directly into the ceiling, no housing required. Canless is usually cheaper in labor; canned is usually cheaper to maintain, because you can swap just the module or trim without opening the ceiling.
I also went to a training session run by a local distributor called Spotlight Academy. They hold seminars on LED technology for facilities people, and honestly, I'd been ignoring their emails for years. The instructor spent a solid half hour on this exact comparison. His rule of thumb: canless is a fine choice if you're comfortable replacing the entire unit when it fails; canned makes more sense when you're in a building you'll own for another decade.
For our office, we went with canned housings and a Cree LED retrofit module. I'll be honest: the canless fixtures were cheaper up front, around $14 per unit versus $24 for housing plus trim, and the labor spread nearly canceled out. But we expect to stay in this building for fifteen more years, and I'd rather my maintenance crew swap a $40 module than hunt down a discontinued $200 complete fixture in 2033. (The storage room already holds a graveyard of orphaned retrofit kits. I did not want to add to it.)
That doesn't mean canned is right for everyone. If you're in a leased space, or under a drop ceiling with easy access, canless is likely the smarter call. The right answer matches the expected life of the building, the ceiling conditions, and your tolerance for future pain.
What the Real Numbers Looked Like
We installed the six Cree LED parking lot light fixtures in late April, using the same contractor. Line items:
- Cree fixtures, six units, delivered: $14,200
- Labor, same contractor: $1,900
- Disposal of failed import units: $270
That's $16,370 for the replacement. Add the $9,610 I burned on the cheap experiment, and the total was $25,710. If I'd bought Cree in October, fixtures plus labor were $16,100. The cheap route cost me about $9,600 extra, plus four months of a half-dark parking lot and more stress than I care to re-live.
On the positive side, the energy reduction is real. The new fixtures draw about 1,050 watts total versus roughly 2,400 watts for the old metal halide units. At our commercial rate of about $0.11/kWh and around 4,100 hours of annual operation, that's roughly $600 per year in electricity savings — before counting the labor we used to spend re-lamping and cleaning the old reflectors. The office retrofit should trim another 30% off interior lighting load.
I also bought a Cree LED flashlight for the maintenance team. It's about $40 and has already survived a drop from an 8-foot ladder, which is more than I can say for the last two cheap flashlights. It's become a running joke that the first Cree LEDs we ever owned were in a flashlight rather than, you know, in the actual fixtures. These days the joke lands a little differently.
If you're reading this because you're in the middle of your own quote comparison, here's the checklist I wish someone had handed me:
- Quote total cost of ownership, not fixture price. Freight, failure rates, replacement labor, disposal, and what the warranty actually requires from you once you read the fine print.
- Ask for the thermal spec. Lumen output tells you how bright a fixture is when new. Driver temperature and L70 data tell you how long it stays bright. In my experience, that's the biggest difference between a $400 fixture and a $700 one.
- Check the warranty logistics. "You pay return freight and wait six weeks" is a different product than "we ship you a replacement first." That difference was our entire tragedy.
- Match the fixture to the building's expected life. Canned vs canless recessed lighting, like most lighting choices, comes down to how long you'll own the problem.
To be fair, the cheap fixtures weren't all bad. They worked fine for the first six weeks, and the light color was pleasant. And the old metal halides ran for fifteen years before they quit. But a permanent facility on a 24-hour schedule needs more than a good first impression.
And Cree fixtures aren't right for every application either. If you don't plan to own the building long, if the mounting pattern doesn't fit your existing poles, or if light spill is a concern, get the photometric file and check it against your layout first. Plenty of good mid-tier fixtures exist. The point was never the name on the box. The point is to run the numbers that include year four, year six, and year eight — not just the invoice date.
I still get a little annoyed when I remember how confident I was last October. But honestly, the experience made our procurement process better. We now require three quotes minimum and a TCO spreadsheet for anything over $2,000, and I've already caught two other purchases that would have repeated this exact pattern. If my story saves you one dark parking lot and a few thousand dollars, it was worth the retelling.