The most expensive lighting mistake I've signed off on wasn't a bad fixture—it was a "compatible" replacement part that cost us $22,000 in a redo. In our Q1 2024 quality audit, we rejected 18% of first deliveries across roughly 200 lighting line items. Almost none of them failed for price or lead time. They failed because the spec sheet matched and the install didn't.
Here's the short version of what I check before approving replacement parts, DTL dark-to-light photocontrols, chandelier assemblies, and track heads for art—and the one situation where I've learned to walk away from a quote entirely.
I'm the quality and brand compliance manager for a commercial facilities group. Since I implemented our verification protocol in 2022, I've reviewed roughly 200 lighting items a year before they reach technicians or go into a client spec package. The protocol exists because of a batch of 400 LED strip connectors that arrived with reversed polarity markings. The vendor claimed the parts were "within industry standard." They weren't—normal keying tolerance is ±0.5 mm, and these were off by 1.2 mm. We rejected the batch, they redid it at their cost, and every contract since has included a first-article inspection clause.
Replacement Parts: "Compatible" Is a Marketing Word, Not a Spec
From the outside, a replacement part looks like a solved problem: match the model number, match the wattage, order it. The reality is that "compatible" covers everything from a true drop-in to a part that needs field modification you didn't budget for.
What I check now, in order:
- Mechanical interface first. Keying, screw spacing, connector housing depth. A 1 mm difference in a twist-lock socket means the fixture doesn't seal, and an unsealed exterior fixture in our climate fails within two seasons.
- Thermal path second. A replacement LED module with the same lumen output but a different heat sink geometry runs hotter. We had one batch of strip connectors where the thermal pad was 0.3 mm thinner than the OEM part. Nothing on the spec sheet flagged it. The first summer, 6% of the run developed flicker.
- Photometric profile third. Same lumens, different beam distribution. On paper, interchangeable. In a parking garage, not even close.
The upside of sourcing Acuity Brands lighting replacement parts through a broad portfolio (controls, fixtures, and parts from one channel) is fewer vendors to chase. The risk is assuming "same brand family" means "same engineering." It doesn't. I kept asking myself on the $22,000 redo: was a 12% unit-cost savings worth a re-commissioning line item? The answer was no, and I've applied that math to two smaller orders since.
To be fair, not every part needs this level of scrutiny. If you're replacing a failed unit with the identical model number from the same production year, you're fine. The checklist matters when you're crossing generations, sourcing from a distributor's "equivalent" list, or specifying for a project you won't personally inspect.
DTL Dark-to-Light Photocontrols: The Threshold Is the Whole Product
A photocontrol is one of those components where the spec sheet tells you almost nothing useful. The datasheet says "dark-to-light switching." What it rarely states clearly: the foot-candle threshold, the time delay, and the hysteresis—the gap between the light level that turns the fixture on and the level that turns it off.
When we specified Acuity Brands' DTL dark-to-light photocontrols for 40 pole-mounted fixtures in 2023, the two candidate datasheets looked identical. What differed was the time delay: 30 seconds vs. 90 seconds. During a thunderstorm that rolled through at 4 p.m., the 30-second units cycled the HID ballasts three times in eight minutes. Two ballasts failed within a month (unfortunately, a month after the warranty inspection closed). The vendor's position was that the units met spec—they did; the spec just didn't include delay. We now write delay and hysteresis into every controls submittal, and I don't approve a photocontrol without a bench test at 1, 3, and 10 foot-candles.
The counterintuitive part: the cheaper photocontrol wasn't the problem. The expensive one with the "adaptive" algorithm was, because the adaptation logic was undocumented and, I think, behaved differently across temperature ranges. More features, less predictability.
Chandeliers (Including the Huge Ones): Gravity Doesn't Read Spec Sheets
Decorative chandeliers—especially the huge chandelier installations over lobbies and stairwells—are where I see the widest gap between what a lighting manufacturer is good at and what the project needs. A controls company can be excellent at DTL, occupancy, and networked dimming and still be the wrong partner for a 400-pound custom glass chandelier. That's a different craft: rigging, structural attachment, cleaning access, and lamp replacement without scaffolding.
"This isn't our strength—here's who does it better." That was a vendor we'd worked with for two years, on a custom chandelier spec. They referred us to a specialist shop. That conversation earned them three other contracts. The supplier who says "we can do anything" is the one I now schedule extra inspection time for.
On huge chandeliers specifically, three checks catch most problems:
- Suspension hardware rated for at least 4x the fixture weight, with a documented attachment method to the structure—not just to the ceiling grid.
- Lamp or LED module access from below or from a catwalk, not from a lift that has to be rented every time a module fails. (Renting a 40-foot lift every 18 months adds up faster than the fixture cost.)
- Finish and cleaning spec that matches the actual cleaning chemicals the facilities team uses. Lacquered brass and ammonia-based glass cleaner are a bad combination, and nobody flags it until year two.
Track Lighting for Art: CRI Is the Floor, Not the Ceiling
"What's the best track lighting for art?" is the question I get most from gallery and institutional clients, and the honest answer is that CRI 90 is the minimum, not the goal. For artwork, I specify against IES TM-30 (the color fidelity and gamut metrics that replaced CRI for serious specification), and I care about three things the datasheet usually buries:
- R9 (deep red) and R12 (blue) values. A fixture can hit CRI 90 overall and still render skin tones and certain blues badly. For art with reds, R9 below 50 is a non-starter.
- UV and IR output. "Art-safe" is a claim, not a standard. I ask for the spectral power distribution and check sub-400 nm output directly.
- Beam angle tolerance. A 15-degree beam that's actually 19 degrees changes the whole lighting design. I've measured ±4 degrees on units sold as ±1 degree.
Per FTC guidance, claims like "gallery-grade" or "art-safe" need substantiation—the Commission's advertising rules require that objective claims be backed by evidence, and "art-safe" is an objective claim. If a vendor can't produce spectral data, that's the answer. (I've had exactly one vendor offer to send a spectroradiometer report without being asked. That's the kind of supplier you keep.)
When This Checklist Is Overkill
I don't run this process on every order. If you're replacing a failed fixture with the identical model from the same production run, or restocking a consumable you've bought for three years, the inspection is a five-minute visual check. The full protocol is for first articles, cross-generation replacements, custom decorative fixtures, and anything going into a space where re-work means scaffolding or a client walkthrough.
Looking back, the decision I'd redo is the 2021 connector order. At the time, the distributor's "equivalent" list seemed reliable, and the price delta was meaningful. If I'd paid for first-article inspection upfront—about $400 on that order—I'd have avoided the $22,000 redo. Granted, that's hindsight talking. But I've stopped treating "compatible" as a specification, and I've started treating a vendor's willingness to say "that's not our strength" as a quality signal, not a red flag.