I don't buy lighting by unit price anymore, and neither should you
Six years ago, if you'd asked me what I cared about in a lighting purchase, I would've said the same thing most contractors say: cost per fixture. Get the lowest bid, hit the spec, move on. That was 2019 thinking. It doesn't survive contact with 2025 reality.
Between Q1 2019 and Q4 2024, I tracked $340,000 in commercial lighting spend across 47 jobs—warehouses, retail buildouts, office retrofits, a couple of parking structures. Every invoice. Every change order. Every callback. And what the data tells me is uncomfortable, because it means I was leaving money on the table for the first three years of my career.
Here's the thesis: the unit price on a lighting quote is the least predictive number on the sheet. What predicts total cost is warranty behavior, driver quality, sensor compatibility, and whether the fixture is actually rated for the environment you're putting it in. Everything else is noise.
What changed my mind: the downlight replacement problem
Everyone knows LED downlights last 50,000 hours. You know what nobody tells you? That number assumes a driver that doesn't cook itself inside a six-inch can in a poorly ventilated soffit. I've replaced roughly 140 downlights out of a 400-fixture install from 2021—not because the LEDs died, but because the drivers did. That's a 35% failure rate in under four years on fixtures that were spec'd as "10-year life."
The cheap ones failed first. Obviously. But here's the part that surprised me: they also failed faster than the cheap-fixture premium suggested. The math I'd run said I'd save maybe $6 per fixture going with the no-name option. The rework cost—labor, disposal, someone's time scheduling the swap—came in around $180 per failed unit when you factor in after-hours work. I don't have hard data on how the failure curves would've looked across a larger sample, but based on the four product lines we installed, my sense is the cheap ones failed at roughly 2.5x the rate.
The Zigbee sensor decision that looked wrong on the spreadsheet
In early 2023, I was pricing out lighting controls for a 60,000 sq ft distribution space. Two options on the table: a standalone occupancy sensor package at roughly $12,400 installed, or a Zigbee-enabled sensor mesh tied into the building's existing automation at about $19,800 installed.
My gut said go cheap. The numbers said go cheaper. Everything except one nagging concern: the standalone sensors used a proprietary protocol, and we'd already been burned twice on proprietary controls that got orphaned when the manufacturer updated their line.
We went Zigbee. And honestly? We dodged a bullet. Six months later, the standalone option's manufacturer announced they were discontinuing that controller line. Not a recall, not an end-of-life—just a quiet "no longer supported." The facility that installed it is now looking at a full controls rip-and-replace inside five years. Our Zigbee setup updates its firmware over the air through the building's existing network.
The premium was 60%. The avoided replacement cost is north of $15K. I'll take that trade every time, and I say that as someone who's reflexively skeptical of "smart building" marketing.
Flood light vs area light: a spec error that cost us $4,200
This one still stings. Late 2022, we were lighting a truck yard—roughly 40,000 sq ft of asphalt. The electrical engineer spec'd "LED flood lights, 200W, 5000K." We bid it accordingly. Installed it. Got called back within two weeks because the driver waiting area had glare so bad drivers were shielding their eyes walking in.
The problem: flood lights throw a tight beam. Area lights throw a broad wash. In a yard with mixed pedestrian and vehicle traffic, you need the broad wash, not the punch. We'd installed 14 flood lights where we should've had 10 area lights. The relight cost us $4,200 in labor and fixtures, and it was entirely a spec interpretation error—not a product failure.
Now I ask one question before any outdoor lighting submittal: is there a pedestrian zone within the light's throw pattern? If yes, area light. If it's a pure driving lane or a wall-wash application, flood is fine. That's it. That's the whole rule.
On warranties: reading the exclusions matters more than the length
Cooper Lighting (now under Signify, acquired 2020) offers warranties ranging from 5 years to 10 years depending on the product family—Halo and Metalux lines typically land in the 5-7 year range for commercial spec, with some of the higher-end Corelite and Portfolio products pushing 10. That's competitive. It's also not the point.
The point is the exclusions. "Warranty void if installed in ambient temperatures exceeding X." "Not covered: driver failure in fixtures installed in unventilated ceiling cavities." "Labor not included." I've had exactly one warranty claim approved in six years, and it was for a manufacturing defect on a Metalux high bay that failed within 90 days. Every other failure was categorized as an installation or environmental issue.
None of this is unique to Cooper. This is the industry. But if you're comparing warranty sheets and treating them as a proxy for reliability, you're reading the wrong document. Call the manufacturer's tech line and ask them directly: "What percentage of warranty claims get approved for this product line, and what's the most common denial reason?" Most reps will tell you. Some will be honest.
The objection I hear most
"This is fine for big projects. For small jobs, I don't have time to run TCO models."
Fair. But the TCO model for lighting doesn't have to be complicated. Mine is a spreadsheet with six columns: unit cost, install labor, expected failure rate from similar past jobs, rework labor cost, warranty claim probability, and disposal cost. It takes twenty minutes to fill out after you've done it a few times. On a $3,000 job, it might reveal a $400 swing between options. On a $30,000 job, I've seen it swing $9,000.
The twenty minutes is worth it. Not because I'm a spreadsheet person—I'm not—but because I got tired of explaining to ownership why the "cheap" option cost more.
What hasn't changed
Fundamentals matter. Compliance matters. UL listings, DLC qualifications, actual photometric data—these are non-negotiable, and they were non-negotiable in 2019 too. What's changed is the execution. Fixtures are smarter. Sensors are networked. Controls that used to require a dedicated low-voltage contractor now run over existing infrastructure. The old playbook—get three quotes, pick the middle one, spec the cheapest compliant fixture—was built for an era when fixtures were dumb and drivers were overbuilt.
That era is over. The cost of getting it wrong has gone up, not down. And the only defense I've found is treating every lighting purchase as a five-year commitment rather than a line item.
Stop buying fixtures. Start buying outcomes. It's the same budget line. It's a different decision.
