PV Module Sourcing Checklist: 7 Steps Before You Sign a Supply Contract

This is for B2B buyers — distributors, EPCs, project developers — who are about to place a module order and want to skip the four or five mistakes that eat margin six months later. It's the checklist I run before I sign off on a supply agreement. Seven steps. If you do it properly it takes about two weeks, and it has saved me more than one awkward phone call with a project owner.

I handle quality and brand compliance for a solar distributor. I review around 200 module SKUs a year across a dozen and a half suppliers. Last year I rejected about 11% of first-delivery samples — mostly small stuff (labeling, packaging, flash data that didn't match the label), but two of them were serious enough to walk away. Here's what I do so that rejection never has to happen.

Step 1: Write the spec before you talk to a single supplier

Not after. Before. The moment you're on a call with a sales rep, the frame gets set by them — their portfolio, their sweet spot, their binning policy. Write your own spec first, in your own words, and put a number on everything.

Minimum items: wattage range and power tolerance (0/+5W is common on most commercial module lines; 0/+3W shows up on higher-binned products), cell technology, module dimensions and weight, junction box and connector type, frame height, fire rating, and the certification set you actually need for your destination market.

The one most buyers miss: flash data tolerance. Ask for the allowed deviation between the nameplate value and the flash test result, in writing. If a supplier won't commit to a number, that tells you something.

Step 2: Check the factory behind the brand — and read the financials

A photovoltaic module is a 25-year product with a 10- or 12-year product warranty. You're not buying watts, you're buying a counterparty. So check the counterparty.

Two things to look at:

• Manufacturing footprint. How many cell and module facilities, where, and which ones are actually running your SKU. A brand with five fabs can serve you from a different one next quarter. Know which one you're contracted to.
• Published results. Listed manufacturers file half-year and annual reports. The H1 2024 filings are still the most recent full-cycle data set as of early 2025 (verify before you rely on it). As an example of what a Tier-1 discloses: Trina Solar's H1 2024 report put module shipments at about 34GW, alongside a full revenue and margin breakdown. Read the revenue line and the gross margin line together — a big revenue number with compressed margin tells you more about your negotiation room than either figure alone.

People assume the higher price causes the better product. Honestly, it runs the other way. The quality system — the incoming cell sorting, the EL re-test, the flash retesting, the traceability database — is what earns the tier, and the tier is what lets them charge more. You're paying for the process, not the logo.

Step 3: Decide OEM vs private label — and decide it on purpose

This is where a lot of buyers get sloppy. The default assumption is that private label is cheaper. Sometimes true, sometimes not once you add it up.

Rough split of what actually changes:

• OEM (branded): you get the manufacturer's warranty paperwork, their traceability system, their name on the label. Your customers can verify the product against the maker's own database. Higher unit price, lower paperwork burden.
• Private label: your brand on the frame and the box. Same line, same cells, different label. Lower unit price, but you own the warranty interface. If a module fails in year six, the customer calls you, not the factory. Your reputation is the one on the roof.

The step people skip: ask what changes about the warranty when you switch the label. On some private-label agreements the manufacturer's 25-year performance warranty still runs to the end buyer. On others, it runs only to you, and you're back-to-back with your customer. Get that in writing before you price anything.

Step 4: Request the full document package, not the brochure

A product datasheet is marketing. What you want is the verification set. Ask for:

• IEC 61215 (design qualification) and IEC 61730 (safety) certificates, with the issuing lab named and the certificate number visible
• Flash test report for your specific production batch, not a generic line
• EL (electroluminescence) images, ideally for a sample of the batch
• BOM disclosure for the main components — cells, encapsulant, backsheet, junction box — even if at a category level
• Factory audit reports if your market requires them

If a supplier sends the brochure and calls it a data package, that's a data point about them. Not necessarily a disqualifier, but you now know how the relationship will go.

Step 5: Sample test on your own bench, not on theirs

The numbers on a supplier's flash report and the numbers on your bench won't always match, and I've learned to care a lot about the gap.

I had a case where every cost comparison said go with a cheaper vendor — 12% under our incumbent, similar spec sheet, decent certifications. Something felt off. Their sample came in with a junction box spec that didn't match the datasheet, and their response time during the sample phase was measured in days, not hours. I stuck with the incumbent. Six months later I heard from a peer that the budget vendor had a batch with a lamination issue that broke down in humid conditions. I never found out whether that was systemic or a one-off. But the responsiveness thing was real, and it was a preview.

What I actually test: flash consistency across the sample (are the five modules within a tight band, or scattered?), EL image quality, frame and glass integrity after a drop test at your warehouse height, connector mating with the cables you already stock. EL under your own light source if you have one. If you don't, borrow one.

Step 6: Put traceability and warranty terms in the PO, not the relationship

Verbal assurances are worthless the quarter the sales rep changes jobs. The purchase order is the document that survives.

Minimum clauses:

• Serial-level traceability back to the production batch and date code
• Defect rate threshold that triggers a batch review (not a full recall — a review)
• Warranty administrator named — who handles a claim in year eight, and where
• Dispute mechanism, including which lab's test result is binding
• Replacement timeline, in days, for confirmed defects

This is where the efficiency argument actually lands, in my experience. Manual paper-based claims handling turns a straightforward defect into a six-week email chain. Suppliers who've digitized their claim intake — a portal, structured data entry, a case number — cut turnaround dramatically. In our own process, moving from emailed forms to a structured intake cut average claim resolution from around 19 days to 8. Not magic. Just fewer places for a form to get lost.

Step 7: Audit the second shipment, not the first

Everyone behaves for the first delivery. The first batch is the audition.

The batch that tells you who you're dealing with is delivery two or three, once the relationship is comfortable and the sales pressure is off. Run the same checks as step 5 on batch two and three. Track the delta. If flash data drifts, if EL images get sloppier, if packaging gets thinner — that's your signal, and it's early enough to fix.

Common mistakes worth naming

Comparing sticker prices across different bins. A 580W module and a 605W module aren't the same product, even if they're the same family. Normalize to $/W and then check the bin tolerance.

Assuming a Tier-1 brand means every SKU is Tier-1 quality. It usually does. Usually. Verify the specific line, not the brand.

Skipping the OEM vs private label warranty question because the price difference was small. The price difference is the small part.

Looking back, the one I'd redo: an early order where I accepted a spec tolerance verbally because the rep seemed straight. Nothing went wrong, and I got lucky. I wouldn't rely on that again. Get it in writing — it costs nothing and it changes the entire conversation when something drifts.

That's the seven. Two weeks of work, and it's the cheapest insurance you'll buy on a 25-year product.