Last March I signed off on 24 recessed lights for the east wing of our main building. Three quotes, cheapest wins, $2,900 installed. I felt fine about it—the old fixtures had been flickering since the previous summer and I was tired of the work orders.
Six weeks later, one went dark. That's when I found out it wasn't a bulb. It was a sealed LED module. The replacement part was $47, the electrician was $85, and because the module was proprietary, we couldn't swap in anything cheaper. By the time the ninth fixture failed in October, we'd probably spent more on modules and callouts than the original install.
That was the surface problem. The real one took me another six months to see, and it wasn't about recessed lights at all.
I thought it was a supplier problem. It was a category problem.
Same fiscal year, same budget, I made two other lighting purchases that went sideways. When I look at all three together I want to go back and shake some sense into the version of me who signed off on them. Here's what happened.
The reception area. We'd specced a sputnik chandelier for the front lobby—something modern, visible from the elevator bank, the kind of thing executives notice on a walkthrough. The one that arrived matched the photo. It also rendered skin tones like everyone had a mild fever. I'd checked the diameter and the finish, because that's what the spec sheet put first. I hadn't thought to check the color rendering index at all.
The board room. This one was worse. A kristal chandelier—modest size, five points, warm look in the catalogue. The mounting plate was 11 inches. Our ceiling junction box was 4 inches. The electrician made it work, but it took a custom adapter plate and a second visit six days later, during which the board room had a hole in the ceiling and a note taped to the door. "Standard size" meant two different things to me and the supplier, and we only discovered the mismatch after the crate was open.
The rooftop greenhouse. Last one, and the one that still bugs me. We keep a small employee garden on the roof—tomatoes, peppers, a few herbs—and the old fixtures were doing basically nothing. I spent two evenings reading gavita pro rs 2400e led review articles, picked a system based on wattage, and installed it. The plants grew tall and thin. The pepper yields were bad. My electricity bill went up about $180 a month and I couldn't explain it to finance, because on paper I'd bought a smaller system than the one it replaced.
Three purchases. Three different failure modes. But they weren't actually different. I was buying all of them with the same mental model.
The mistake: treating "lighting" as one purchase category
Here's what I finally understood. Decorative lighting, functional lighting, and horticultural lighting are not variations of the same purchase. They're different product classes with different evaluation criteria, and in some cases different physics. You wouldn't buy accounting software and warehouse software with the same checklist, and you shouldn't buy a lobby chandelier and a greenhouse fixture with the same one either.
The reason this is easy to miss is structural. They all get lumped into the same line item on a budget spreadsheet, they all get quoted in dollars per unit, and they all show up in the same "lighting and electrical" vendor catalogues. The category confusion isn't in the products. It's in the procurement process I was using.
Decorative fixtures: sputnik chandelier, kristal chandelier, and everything visible
What matters here is visual, not technical. Color temperature (typically 2700K–3000K for a warm look, 4000K for a modern neutral), color rendering index (CRI 90+ is the difference between "warm" and "clinical" on a client's face), and the physical match to the space—diameter, drop height, mounting plate dimensions, junction box compat.
The review trap is worst in this category. A sputnik chandelier might have 4.8 stars from 500 buyers. None of them have your ceiling height, your photographs on the wall, or your client walking through the lobby at 9 a.m. on a Monday. Reviews tell you the fixture arrived intact. They don't tell you it'll match your building.
Functional lighting: can you change the bulb in recessed lighting?
This is the first question, and it should be answered before anything else gets discussed. Not all recessed lighting has a replaceable lamp. Sealed LED modules are common, increasingly so, and they carry a hidden cost structure: cheaper upfront, much more expensive over ten years, because when a module fails you're replacing the whole fixture—or paying for a proprietary part that only one supplier makes.
The other numbers worth pulling are the datasheet values: LM-79 photometric reports and LM-80 lumen maintenance data. Both come from Illuminating Engineering Society (IES) testing standards and are the closest thing to a fair comparison across suppliers. If a supplier can't produce either, that's information too (note to self: I should have treated the missing LM-79 as a red flag the first time). For a functional purchase, the question isn't "what's the cost per fixture?" It's "what's the cost per fixture over ten years, including replacements, labor, and downtime?" On that math, our sealed modules turned a $2,900 install into roughly $7,400 across sixteen months.
Horticultural lighting: gavita under canopy lights, top lights, and everything else
This category is where the evaluation criteria are the most technical and the least intuitive for someone coming from general facilities buying. Wattage means almost nothing. What matters is photosynthetically active radiation, measured as PPF (photosynthetic photon flux, in µmol/s) and delivered as PPFD at the canopy (µmol/m²/s).
For our situation—low ceiling in a rooftop greenhouse, mixed canopy heights, tomatoes and peppers—we needed two things I hadn't thought about: a real light distribution map at our actual mounting height, and under-canopy fixtures to get light to the lower leaves where the fruit actually sets. That's why gavita under canopy lights kept coming up in the supplier conversations. It wasn't brand loyalty. It was that overhead light at 8 feet of canopy distance mostly hits the top leaves and not much else.
The DLC Horticultural Lighting Technical Requirements list is a reasonable starting point for verifying claimed performance. If a fixture is on it, the PPF figures have been measured to a standard. If it isn't, the numbers on the marketing page could be measured, estimated, or invented, and there's no way to tell from the outside.
What the mistakes cost, and what I'd do differently
Recessed lights: $2,900 originally, roughly $1,400 in module replacements and electrician visits, and $3,100 for the fixtures we should have bought in the first place. Sixteen months, two purchases, one working wing.
Reception chandelier: one re-spec, one return shipping charge we ate because the vendor's window had closed, and about six weeks of a lobby with the "temporary" fixture still hanging. The CRI 90 replacement added $1,900 to the line item, because it wasn't the cheapest tier. Looking back, I should have paid the premium the first time. At the time, the difference looked like 40% of the budget line, not 4% of the total facility spend.
Greenhouse: about $2,100 in wasted electricity over nine months, plus a growing season I'd rather not talk about. The new lights cost 2.4x per fixture versus the ones I originally bought. They used less power per unit of light delivered, and the yield difference was visible in the first month. That $2,100 number is what I now point to when someone argues for the cheaper fixture on a spec sheet.
Total, across the three: a shade under $9,000 I could have spent better, plus about 40 hours of my own time on returns, calls, re-quotes, and holding the electrician's schedule together. The failures weren't caused by bad suppliers. All three suppliers were fine. I bought the wrong product because I asked the wrong question.
How I buy lights now
Three separate templates, one per category, and a hard rule that they don't get merged. I keep them in a shared folder and the first thing I do with a new quote is figure out which template it belongs to.
For decorative fixtures, the space gets photographed with a color card, and the spec sheet has to include CRI and CCT before anything else gets discussed. Reviews don't enter the process until after the shortlist is final.
For functional lighting, the first question on every quote is whether the light source is replaceable. If the answer is sealed, the number that matters is the ten-year cost, not the per-unit price. I also started asking for the LM-79 report as a gating item, not a nice-to-have.
For horticultural lighting, I don't start with watts or price. I start with a PPFD map at my actual mounting height, then work backward to fixture count. Under-canopy goes into the plan from the beginning, not as a second wave after the first install disappoints.
My view on lighting pricing, after this year: the cheapest option has cost us more, once, than any of the "expensive" ones did. That's not a general rule and I've been the guy defending the budget quote for long enough to know it. But it's been true often enough at this company that I'll now argue for a 30% premium on a verifiable spec sheet without flinching.
The other habit I've dropped is buying on reviews. A gavita pro rs 2400e led review from someone with a 12-foot ceiling doesn't help me at 8 feet. A five-star recessed light review doesn't tell me whether the module is sealed. Reviews are good for one question—"did it arrive broken"—and pretty much nothing else.
Last month the east wing finally got the fixtures we should have bought in March. Replaceable lamps, standard junction box, LM-79 on file. Nobody's noticed the change. That's the goal.
