2026-09-29

How to Evaluate Solar Module Manufacturers: Seven Mistakes and $180,000 Later

A PV procurement post-mortem: why comparing $/W on datasheets keeps failing buyers, what the spec sheet actually hides, and the eight questions I now ask every solar module manufacturer before I sign anything.

In my first year buying modules (that was 2016), I assumed the module decision was a math problem. Collect the datasheets, normalize everything to $/W, pick the lowest number with the shortest lead time. I built a comparison model with 14 tabs for a 2.1 MW order. I was extremely proud of that model. It was also the reason we ate roughly $11,000 in re-engineering costs six months later.

Nine years and seven documented mistakes on, that's about $180,000 in wasted budget if you want the ugly number. I still keep a version of the spreadsheet. I just ask eight more questions before I open it, and none of those questions are answered anywhere on a photovoltaic module specification sheet.

The Problem You Think You Have

Almost every buyer I talk to believes the hard part is finding the module. Sourcing. Getting quotes. Negotiating two cents off per watt.

It isn't. The hard part is that "best module" isn't a property of the module. It's a property of the relationship, the paperwork, and the bill of materials (BOM) you didn't lock down.

I found that out slowly.

Why the Spec Sheet Keeps Letting You Down

You're comparing numbers measured somewhere you'll never be

Under IEC 60904-3, standard test conditions are defined as 1,000 W/m² irradiance, 25°C cell temperature, and an AM1.5 reference spectrum. A rooftop in July is none of those three things.

So when you line up a 580W pv module against a 575W module and declare a winner, you're comparing laboratory snapshots. The number that actually decides your yield is the temperature coefficient, and it's printed in a footnote that most buyers — including me for about three years — scroll straight past.

Here's the one that stung. In 2019 I compared two modules at nearly identical wattage and argued hard over a two-cent-per-watt difference. One had a temperature coefficient around -0.29%/°C. The other was nearer -0.35%/°C. On a hot roof, that gap is worth more than the price difference we spent three weeks fighting about. I never calculated it. Nobody on my team did either. We learned it from the customer's asset manager about two years later, in a very polite and very painful email.

A datasheet is also a snapshot in time, not a contract.

A type approval is not a promise

IEC 61215 and IEC 61730 qualify a design, tested on a specific sample, with a specific BOM. Change the cell supplier, swap the encapsulant, move to a different junction box, and you've changed the product. There's a reason the industry has a separate standard — IEC TS 62915 — specifically for retesting after design changes.

Ask a supplier what triggers a retest, and who pays for it. Roughly half the ones I've asked go quiet for a day or two. That pause is information.

The logo on the frame isn't always the company on the hook

This one matters most if you're buying OEM or private label. Your name goes on the module. That's the easy part. The question is whose name is on the warranty obligation, whether that legal entity has existed for more than a few years, and whether a lender will accept it.

I once spent a pleasant afternoon reviewing a very attractive private label quote, then asked who signs the 12-year product warranty. The answer was a trading company incorporated eight months earlier. We walked. Two years later that trading company was gone, though to be fair, the modules themselves were probably fine. Probably isn't a word that belongs in a supply agreement.

The Bill That Never Shows Up in the Spreadsheet

The September 2022 shipment is the one I still use in training.

1,840 modules, 550W nominal, 0/+5W positive-tolerance binning. First production lot flashed between 552W and 555W. Second lot flashed at 550W and 551W. Both lots were labeled 550W. Both were compliant. Both met spec.

Put them on the same string and the mismatch cost us around 2.4% of the array's output on that section. The modules weren't defective. They were correct, which is exactly what made it expensive — there was nothing to claim, no RMA, no credit note, no conversation to have with anyone. Just a yield number that came in under the model, forever.

The same thing happened with delivery. The numbers said go with the supplier who was three cents cheaper per watt. My gut said the three-week silence during the quote process was a preview of something. I went with the numbers. The shipment arrived eleven days late and we re-sequenced the install crew twice.

And people still ask me why I care about flash test data and EL images. It's because a 3-week delay on a financed project costs more than the entire $/W difference between tier-1 and tier-2 on a small commercial rooftop. The module line item is often the cheapest thing in the budget and the most expensive thing to get wrong.

(Verify current pricing before you quote anyone. Module pricing moves monthly, and anything I tell you from last quarter may already be stale.)

The Actual Root Cause

It took me about four years and something like 150 orders to understand that I was evaluating a manufacturer when I should have been evaluating a supply relationship.

Granted, that sounds like consultant talk. Let me make it concrete. A module is close to a commodity. Two tier-1 modules with the same cell technology, similar BOM, and comparable certifications will perform roughly the same in year one. The differentiation shows up in year five and year eight — in whether someone answers the phone, whether the BOM stayed the same, whether the warranty entity still exists, and whether spare modules are sitting in a warehouse near your project or on a ship somewhere.

What counted as best practice in 2018 — collect three quotes, check for IEC certificates, negotiate hard — isn't enough in 2025. The technology matured. The failure modes moved. Ten years ago the risk was 'will this module work.' Now the risk is 'will this documentation survive scrutiny.' The fundamentals haven't changed; the execution has completely transformed.

What I Ask Now (Abbreviated)

Eight questions, asked before price is even discussed:

  • Who is the legal entity on the product warranty, and how long has that entity existed?
  • Can the BOM be locked into the supply agreement — cell supplier, encapsulant, glass, junction box?
  • Will you provide flash test data for the specific production lot, plus EL images?
  • What is the temperature coefficient tolerance, not just the typical value?
  • Which revision of IEC 61215 and IEC 61730, and what's your policy under IEC TS 62915 for design changes?
  • What changes trigger a retest, and who pays for it?
  • For OEM and private label: who owns the RMA process, and where is the spare module inventory physically held?
  • Will you hand over your own PID and LeTID test data, or point me to third-party results?

Companies that can answer all eight exist. I've had direct, documented answers on spec tolerances and BOM declarations from Astronergy's B2B team on the OEM and private label side, and I've had smaller suppliers stall for three weeks on the same questions. The stall is the answer. You just have to be willing to hear it.

For third-party corroboration, I lean on the PVEL Product Qualification Program and its annual Module Reliability Scorecard. It's not a substitute for your own diligence, but it filters out the suppliers who can't survive extended stress testing.

The Trade I Made

The checklist didn't make me faster. It made me slower. We've lost orders because I asked for lot-level flash data and the competing buyer didn't, and I have genuinely mixed feelings about that.

On one hand, slower quotes cost deals. On the other, we've caught 47 potential issues in the past 18 months using this list — everything from a mismatched connector rating to a warranty entity that couldn't be verified. The best part of finally systematizing it isn't the money saved. It's not lying awake at 3am wondering whether a container is going to clear customs with the wrong junction box on it.

Slower quotes. Fewer surprises. That's the whole trade, and after nine years I'll take it.

Pricing, lead times, and certification requirements vary by market and change frequently. Verify current data with the manufacturer and confirm applicable standards at iec.ch before committing to a supply agreement.


Written by Isabel Moreno