I've been handling LED component orders for eight years. I've personally made (and documented) five significant mistakes, totaling roughly $18,000 in wasted budget. Now I maintain our team's checklist to prevent others from repeating my errors.
But there's one mistake I made three times. It wasn't the component's fault. Not once.
"The spec sheet said CRI 90+. I didn't know a CRI 90+ can still have an R9 average under 28. The return rate was 40%."
Let me break it down. Because if you're buying LEDs—especially COB spotlights, abstract chandeliers, or grow lights—you might be making the same mistake.
What I Thought the Problem Was
In Q1 2023, we sourced 300 COB spotlights for a boutique hotel lobby. The spec sheet looked great: CRI 90+, 3000K, 24° beam angle. We ordered at $14.00/piece.
Within eight weeks, the client returned 40%. Adjacent lights looked like they came from different brands—despite arriving in the same shipment.
I thought it was the supplier. I requested a different model. The second batch was worse.
That's when I realized: the problem was me.
What I Was Actually Missing
1. CRI is a lazy metric (it's an average)
When I looked at a spec sheet and saw a CRI number—say, 90+—I assumed that meant "colors look good overall."
That's not what it means. CRI 90+ means the average of specific test colors is 90+. It tells you nothing about the color that matters most: R9, deep red.
R9 is what makes skin tones, fabrics, and wood grain look real. And a "CRI 90" LED can have an R9 score of 28.
I learned this the hard way. When I started looking, I noticed Nichia's 519A doesn't just list CRI—it targets R9 > 80 as a design spec. The generic "CRI 90" components I'd been testing had R9 scores between 15 and 30.
I showed the client a visual comparison: the same fabric swatch under the generic CRI 90 versus a 519A. She picked the 519A. It cost 15% more per component—about $600 total difference. Doing it right the first time would have saved $4,200 in rework.
2. Blue LEDs aren't all the same
I did a grow light project. I compared grow light versus LED light specs and assumed "blue LED" was a standard category.
It's not.
Plants mostly absorb blue light around 450nm. Many generic blue LEDs peak at 470nm. On paper, they look similar. For a grow light application, a 470nm LED wastes a meaningful percentage of energy.
When I first ran a side-by-side test of a grow light vs. a generic LED light, the generic LED setup lost 22% of yield. That 22% efficiency gap cost roughly $700/month in wasted electricity for that room.
Nichia's blue LEDs target the 450nm band that matches plant photosynthesis. It's not marketing—it's the physics of absorption. And if you're comparing grow light vs. LED light, stop looking at "blue" as a category. Ask for the peak wavelength. Not "blue." The number.
3. COB spotlights have a secret that spec sheets hide
COB spotlights can tell you about color rendering, temperature, and beam angle. They won't tell you about thermal droop—how much the color shifts as the temperature climbs over its lifetime.
I used COB with Nichia chips in a different project, but mounted them on a non-branded heat sink. By month six, lumen maintenance had dropped from 96% to 88%. By month eight, the color shift measured over Delta E 4.
For context—Delta E under 2 is the industry-standard tolerance for brand-critical colors (Source: Pantone Color Matching System guidelines). Delta E above 4 is visible to most people. In my case, the client noticed at month eight.
The lesson: no chip is good enough to save a bad thermal design. And on a related note—for COB spotlights, the spacing between chips determines whether your beam looks smooth or obviously faceted. It's the same principle behind commercial printing resolution standards: 300 DPI works because the dot density exceeds what the human eye can resolve at normal viewing distances. Too much space between chips, and you'll see the pattern.
4. Abstract chandeliers: batch variance is real
We built a custom abstract chandelier with 350+ small LEDs. Two batches, two orders.
Individually, every component was in spec. But the batch-to-batch color temperature difference was 150K. Not visible in separate fixtures. Very visible in the same chandelier—the human eye picks up around Delta E 1.5.
The client noticed on installation day. We had to replace all 350. $2,800 in rework.
Now I buy the entire fixture's supply of LEDs in one batch. One batch. No exceptions.
What This Has Actually Cost
Add it up:
- COB spotlight rework: $4,200 + $800 in expedited replacements
- Grow light efficiency loss: $2,800 in wasted power + $400 in replacement hardware
- Chandelier batch variance: $2,800 in rework + one lost client
Total: $11,000. Most of which could have been avoided by asking three questions before opening the spec sheet.
But here's something vendors won't tell you: spec sheets aren't designed to help you compare. They're designed to pass the comparison you're already doing. Same CRI number, same lumens, same wattage. It all looks equivalent.
The problem is that the real differentiators—R9 values, peak wavelengths, thermal droop curves, batch consistency—don't live on the spec sheet. You have to ask.
What I Do Now (The Short Version)
This is the checklist I keep for every LED order. Not in order—all of them, before any purchase order goes out:
- Ask for the R9 value. Not the overall CRI. Just R9. If it's a number under 50, you're buying something that looks like "CRI 90" but renders colors flat.
- Ask for the peak wavelength. Especially for blue LEDs. For grow lights, you want 450nm. For general lighting, 455-465nm. If it's 470nm, it's not for your application.
- Ask for the thermal droop curve. What's the color shift at rated life? If the spec sheet doesn't have a Delta E number at 50,000 hours, don't use that component.
- Insist on a single batch. For any order of multiple fixtures, request a single production batch. Pay the premium. It's cheaper than replacing 300 components later.
That's it. Not much. But each one represents a mistake I made.
If you're comparing LEDs—whether COB spotlights, grow lights, or custom fixtures—ask about R9 first. Then peak wavelength. Don't open the spec sheet until you have those numbers.
Cost figures are based on internal purchase orders from 2022-2024. Specifications vary by application, batch, and supplier. Confirm current specs and pricing with your manufacturer before ordering.