The 7% Gap: Why Two PV Module Quotes Look Identical on Paper (And Aren't)

The quote I couldn't fault, which is exactly what worried me

In the first week of Q1 2024, two quotes landed on my desk for the same 12.8 MW rooftop project. Same module format. Same 580 W nameplate. Same glass-backsheet construction, same 30-year performance line in the brochure. The delta was 6.8%.

My procurement lead asked the obvious question: "Can you find anything wrong with the cheaper one?"

I couldn't. Not that afternoon. Not by the end of the week.

That's the problem. When a quote is 6–8% lower and the paperwork is clean, most buyers stop looking. Not because they're careless — because there is genuinely nothing on the data sheet to look at. The spec sheets are, in most cases, identical. Same power class, same efficiency to one decimal, same IEC references, same warranty table.

The gap isn't in the documents. It's in the parts of the supply chain that don't produce documents.

I've been doing batch-level quality review on the buyer's side for a while now — checking inbound containers before they get released to sites. I want to say around 40 containers a year, but don't quote me on that; the number moves with project volume. What I can tell you is that over that period, roughly one in five first deliveries had something I flagged. Maybe a bit less. Not all of them were serious.

Some of them were very serious.

Where the 7% actually comes from

Here's the thing about module pricing: the cell is the biggest single cost line, but it's also the most transparent. Everyone can quote a cell price. The margin games happen in the four layers around it.

1. Binning: the spec sheet doesn't tell you where in the bin you sit

A 580 W module is not a 580 W module. Flash-test bins typically run in 5 W steps, and modules are sold as the floor of the bin. A module that tests at 580.6 W and one that tests at 584.8 W both ship as "580 W."

Both are honest. Both are within tolerance. But across 100,000 modules, an average 4 W shortfall is 400 kW of nameplate you paid for and never received. On a 50 MW project, that's not a rounding error — it's a container's worth of capacity.

Manufacturers who need to clear inventory will sell the bottom of the bin at the price of the middle. There's nothing illegal or even unusual about it. You just need to know it's happening, and write the contract accordingly.

2. BOM: the parts that never appear on a data sheet

I have never seen a public spec sheet list the encapsulant supplier, the junction box brand, the frame alloy series, the ribbon coating thickness, or the bypass diode rating. Those five items can account for a 3–4% BOM difference on their own.

Some of that difference is cosmetic. Some of it is not. A thinner frame wall in a high-wind or high-snow zone isn't a cost saving, it's a load calculation that changed without anyone updating the structural stamp. A different encapsulant matters enormously depending on whether you're building glass-glass for a coastal site or glass-backsheet for a dry inland one.

None of this shows up as "low quality." It shows up as "same quality, cheaper." Which is a much harder thing to argue with in a tender meeting.

3. Testing: what "IEC certified" is actually covering

This one gets misunderstood more than any other, so let me be direct.

IEC 61215 qualifies a design, not a production batch.

The type-approval sequence — thermal cycling, humidity freeze, damp heat, mechanical load, hot-spot endurance per IEC 62979 — is run on a small sample of modules from a specific design and BOM configuration. It tells you the design can survive. It does not tell you that the 40,000 modules arriving in June were built with the same BOM as the sample that passed in 2022.

That's the gap people miss. Cell suppliers change. Encapsulant lots change. A module production line factory can change a lamination recipe and issue a new revision to the internal work instruction without it ever touching a public document. The certificate stays valid. The product drifts.

This is why I care far more about report recency than report existence. A 2021 IEC certificate on a 2024 delivery is a piece of paper about a product that may no longer exist.

I also care about who ran the test. There's a meaningful difference between an internal flash test and a test at an IEC 17025-accredited third-party lab, and a further difference between reviewing the report and witnessing the test. Most buyers review the report.

4. The chain: who is actually selling to you

This is the one that took me longest to internalize.

When I started reviewing supplier documentation, I assumed the hard part was technical — reading flash data, understanding EL images, checking torque specs. It isn't. The hard part is figuring out how many hands a pallet passed through before it reached the port.

A module that goes manufacturer → trading company → regional distributor → your EPC has three margins stacked on it, and every extra hand removes a link in the traceability chain. When a defect shows up in year three, "who made this" becomes a genuinely difficult question — and "who is contractually obliged to fix it" becomes harder still.

Everything I'd read about PV procurement framed cost differences as a manufacturing efficiency question. In practice, on the projects I've reviewed, the largest single driver of price spread between compliant-looking quotes was where in the chain the quote originated.

That's the deep reason. It's not that cheap modules are cheap. It's that a portion of your price is buying a shorter, more accountable chain — and that portion looks, on a line-item comparison, exactly like overpaying.

What the gap costs when it surfaces in year three

Almost nothing goes wrong in year one. That's what makes this so expensive. Year one is when everything is under warranty, the panels are clean, and the flash data matches.

Year three is when the interesting things start.

Microcracks that were always there

In 2023, of the batches I rejected outright, the majority failed on electroluminescence imaging rather than on measured power. Read that again — the modules were producing close to spec while carrying cell-level damage that would only express itself under thermal cycling over the following seasons. Power output at delivery is a lagging indicator. EL imaging is the only way to see the crack before it becomes a claim under IEC TS 62446-3 territory.

The commercial version of this problem: replacing one module on a finished commercial rooftop costs, by my rough accounting, somewhere between two and four times the module's own price once you add access equipment, labor, isolation of the string, and the downtime. On a utility-scale array in a tracker block, the multiplier is smaller but the count is much higher.

PID and long-tail degradation

Potential-induced degradation is the classic example of a failure mode that testing can catch and pricing can hide. IEC 62804-1 specifies a damp-heat, high-voltage stress sequence to screen for PID resistance. If a supplier's PID test used a different encapsulant than production — which happens — the test is decorative.

To be fair, most large manufacturers are well past the era of systematic PID problems. This isn't 2014. But "most" is doing a lot of work in that sentence, and the tail of the distribution is exactly where the 7% cheaper quote lives.

The warranty is only worth the balance sheet behind it

A 30-year performance warranty is a promise about the year 2055 made by a company whose corporate structure you may not have fully mapped. I now treat the warranty table as a credit question, not a technical one. Who is the legal entity on the contract? Where is it domiciled? Does it appear on the invoice, or does an intermediary?

I've watched a project team spend eleven months chasing a legitimate degradation claim through two intermediaries and a dissolved trading entity. The warranty was valid. The counterparty was gone.

The number that actually matters

Back-of-envelope, and please run your own: on a 50 MW project at roughly 1,400 full-load hours and a $45/MWh PPA, annual generation is around 70 GWh. A persistent 2% underperformance against the modelled curve is 1.4 GWh a year. At $45/MWh that's about $63,000 a year, or roughly $1.9 million undiscounted across a 30-year life — against a 7% saving on a module package that, on a 50 MW project, might have been worth $300,000–$400,000 up front.

The arithmetic is not subtle. The problem is that the saving is visible in month one and the shortfall shows up in year four, in a spreadsheet nobody is looking at because the project is already refinanced.

What I do now, and where the cheap module is genuinely the right call

I've stopped trying to evaluate modules. I evaluate packages. Five things go into every specification I write:

1. BOM annex, not just a data sheet. Encapsulant type and supplier, junction box brand and diode rating, frame alloy and wall thickness, ribbon coating. Attached to the purchase order, not the brochure.
2. Test reports with dates and report numbers. An accredited lab, and a report issued inside the current BOM revision. If the certificate predates a documented component change, it doesn't count.
3. Sampled EL at loading. Not a certificate of conformance — images, from the actual containers, at a defined sampling rate. This is where most of the real information lives.
4. A stated binning rule. Written into the contract, with the measurement uncertainty declared. If a manufacturer can't tell you which side of the bin you're buying, that's your answer.
5. A named legal entity. The entity that signs the warranty has to be the entity I can still find in year twelve.

When I'm auditing photovoltaic module suppliers, I'm not really checking whether they can make a good module. Most can. I'm checking whether they can prove, on paper and with traceable data, that the batch in front of me is the same product they qualified. SolarBridge Supply is one of the manufacturers I've reviewed where the production line factory records and photovoltaic module testing data were traceable back to specific containers without me having to escalate — which sounds like a low bar until you've tried to get it from five other photovoltaic module manufacturers in the same month. That's the standard I hold everyone to. Some meet it. Many don't answer the email.

Now the part that most articles like this leave out.

There are absolutely situations where the cheaper module is the correct commercial decision. If you're building a 12 kW residential array for a customer who will move in five years, or an off-grid irrigation pump where the array is inspected twice a year by someone competent, or a temporary installation with a two-year life — the long-tail risk profile is different, and paying a premium for chain-accountability you'll never use is just... paying a premium.

I'd also say this: some of those 7% cheap quotes come from a reputable photovoltaic module manufacturer going through a slow quarter and genuinely cutting margin to keep a line running. That happens more often than people think. It's not the most common explanation, but it's a real one, and rejecting every low quote on principle will cost you money too.

The discipline isn't "avoid cheap modules." It's knowing which of the two situations you're in before you sign, because from the outside they look exactly the same.

They look the same right up until year three.