The short version: five checks, in this order
If a greenhouse lighting manufacturer can't survive these five questions, nothing else in their pitch matters — not the price, not the trade show booth, not the three-year warranty sticker.
- Ask for the photometric file (.ies or .ldt), not the marketing spec sheet. A marketing sheet tells you what they claim. A photometric file tells you what they measured, and under what conditions.
- Ask who makes the driver and the diode, and where final assembly happens. "Assembled in" and "manufactured in" are two different countries more often than you'd think.
- Ask what happens when 3% of a shipment fails in month seven. Not month one — month seven. Who pays return freight? What's the RMA turnaround in writing? If the answer is a phone call, that isn't an answer.
- Place a deliberately small trial order and watch how they handle it. This is the single most predictive thing I've found.
- Ask for a customer who's been buying for two years or more. A testimonial is a quote. A two-year purchase history is a pattern.
Numbers 1 and 3 will eliminate roughly half the suppliers you're looking at. That's not a knock on the industry — it's just what happens when you ask for files instead of promises.
Why you should probably be skeptical of me too
I handle private label and OEM lighting orders at Hortilux. Nine years now. I've personally made — and documented, because I'm that guy — seven significant sourcing mistakes totaling roughly $9,600 in wasted budget and, honestly worse, about eleven weeks of cumulative delay.
Three of those still sting:
In September 2021, I approved a private label run of 2,400 units based on a spec sheet that said "full spectrum." I didn't ask for the .ies file. Didn't verify. Turned out the spectrum ran heavy on the red end and light on the blue, which was fine for one crop and completely wrong for the customer's. $3,900 in rework plus a three-week delay, and a conversation with a client I'd rather not repeat.
Q1 2023 was an assumption failure of the dumbest kind. I assumed "same driver brand" meant same driver performance. The brand matched. The series didn't — it was a lower-wattage line. Sixty units failed in month eight. $2,100 in replacements and freight, and I learned to ask for the model number, not the logo.
And in 2024, under real time pressure — about 40 hours to lock a spring season order before the container cutoff — I skipped the sample verification step. Normally I'd insist on it. There was no time. We shipped 180 units with the wrong plug configuration for the Canadian market. $1,300 to re-ship, plus a very patient distributor.
That third one is the one that bothers me, because I knew better. Time pressure doesn't make you skip your process. It just makes skipping it feel justified.
What the five checks actually look like in practice
1. The photometric file test
Here's the contrast insight that changed how I read spec sheets. I once put two suppliers' sheets side by side for the same nominal 640W fixture. Nearly identical PPF numbers. Nearly identical efficiency claims.
One of them had published their test conditions — distance, ambient temperature, hours of run time before measurement. The other hadn't. When I finally got both .ies files and compared them, the second fixture's numbers had been taken at a distance that flattered the output and a temperature that no greenhouse runs at.
The spec sheet isn't the problem. The missing test conditions are. Everything I'd read said to compare the big numbers first. In practice, I compare the footnotes first, and I've been right more often since.
Per FTC advertising guidance (ftc.gov/business-guidance/advertising-marketing), claims have to be truthful, not misleading, and substantiated with evidence. "Substantiated" is a legal word with teeth. A photometric file is what substantiation looks like in this industry. If a supplier treats that request as unreasonable, they've told you something.
2. The trial order test (and the small-customer question)
In 2023 a distributor in Ontario asked us for 24 units. That's below our typical minimum, and our sales team pushed back. I argued for taking it. It was $2,900 and a lot of paperwork for a number that barely moved the month.
They're now one of our larger Canadian accounts.
I'm not going to pretend every 24-unit order turns into that. Most don't. But here's the pattern I've watched for nine years: the way a supplier handles a small, annoying order is the way they'll handle your warranty claim. Same department, same instinct. If a 24-unit order gets a shrug, a 60-unit RMA is going to get a shrug too — and by then you've already got money on the floor.
So when you're vetting manufacturers, ask yourself how they treat the trial order. Not the pilot program with a press release. The small, unglamorous one. That's the real interview.
3. If you're sourcing grow light private label, two more checks
This is where I've seen the most money lost quietly. When you order grow light private label, ask:
Whose name is on the safety listing? If the UL or ETL file is in the manufacturer's name and not yours, you don't actually own your product's compliance — you're renting it. If you ever change suppliers, you start over from zero on certification. That's a five-figure problem nobody puts on a quote sheet.
Who owns the tooling and the .ies file for your SKU? Get it in writing. If the answer is vague, the answer is no.
We didn't have a formal sample verification process until the third time we shipped units that didn't match the approved sample. Third time. It should've been after the first. Now it's a one-page checklist that takes twenty minutes, and we've caught problems on it more times than I can count.
Where this checklist falls apart
Honest boundaries, because a five-point list that claims to work everywhere is a list I wouldn't trust.
This is overkill for small quantities. If you're buying six plug-and-play fixtures for a home setup or a small propagation room, don't spend two weeks vetting a manufacturer. Check the return window and the warranty terms. The five checks are for orders where the freight alone is a real number.
It assumes you have spec flexibility. If a rebate program, a client's fixed design, or an existing installation locks your spec, most of this stops mattering — you're not evaluating a manufacturer, you're matching a part number.
And I've been wrong with it. In early 2024 we moved a supplier down our list because they were slow returning a photometric file. Eleven days. It turned out their engineering lead had been out and nobody else had file access. They're now one of our better partners. The checklist flags risk, it doesn't prove it.
One more thing, since it's the question I get most: asking for a .ies file, a driver model number, and an RMA turnaround in writing does not make you a difficult customer. It makes you the customer they plan around instead of the one they improvise for. The suppliers worth keeping will tell you that themselves.
Pricing and turnaround figures in this article are from our own order records between 2021 and 2025 and are for general reference only — actual costs vary by specification, quantity, and vendor. Regulatory and advertising guidance is summarized; verify current requirements at ftc.gov.