I'm a procurement manager at a 140-person electric utility vehicle manufacturer. I've managed our battery and energy storage budget — roughly $1.9M a year — for six years, negotiated with 30-plus cell and pack vendors, and logged every order in our cost tracking system. This is the checklist I actually run, not the one I'd show a consultant.
Nine steps, from "we need batteries" to a signed spec and a purchase order. It works the same for prismatic LiFePO4 cells, NMC modules, EV battery packs, electric truck battery packs, and home energy storage batteries, because none of these steps are about chemistry. They're about paperwork, samples, and the stuff that shows up in month fifteen.
Read it in order. Skipping step 1 makes steps 5 through 9 meaningless.
The 9-step checklist
Step 1: Write the duty profile before you talk to anyone
One page, one table: voltage window, continuous and peak C-rate, duty cycle, ambient temperature range, cycle life target with a definition, space and weight limits, expected annual volume, destination country, and how the pack gets serviced after it's installed.
Vendors quote exactly what you ask for. If your request is vague, you get a datasheet-optimal quote for the wrong product — and it will be the cheapest quote in your inbox. That's not a coincidence.
Step 2: Read the cell datasheet, not the pack datasheet
Ask for the specific cell model number, chemistry, capacity at a stated C-rate, internal resistance, and cycle life at a stated depth of discharge and temperature. "6,000 cycles" by itself is a marketing number. Six thousand cycles at what DoD? At what C-rate? At 25°C or at 45°C? Ending at 80% of initial capacity or 60%?
Ask those four questions in writing and roughly half the quotes get revised. That tells you something.
Step 3: Get the actual certificate PDFs — and match the model numbers
For lithium cells and packs you should be asking for a UN38.3 test summary, IEC 62133-2 for the cell and battery, UL 1642 or UL 2054 for the pack, UL 1973 if it's an ESS battery, an up-to-date MSDS, and CE plus the EU battery documentation if you ship into Europe. As of 2025 these are still the baseline; verify current requirements with the issuing lab and your freight forwarder, because the rules move.
Here's the thing: certificates often cover a different cell, a different capacity, or a different factory than the one in your quote. Match the model number on the certificate to the model number on the quote to the model number on the shipping label. I've caught a mismatch twice, both times on the paperwork, not the product. Suppliers that publish cell model numbers and test summaries upfront — MSCE does this on its module pages — are simply faster to evaluate.
Step 4: Buy samples. Test them yourself.
Five to ten units. Run a capacity test at your actual C-rate, measure internal resistance, check self-discharge over two weeks, and put a partial cycle test on the calendar. Write the numbers down in a spreadsheet you keep.
I don't have hard data on industry-wide sample failure rates, but based on our last several years of sample orders, my sense is roughly one in five comes in below claimed capacity at the rated C-rate. Not catastrophically. Just enough that the quote was optimistic.
Step 5: Cost the landing, not the unit price
Landed cost per kWh = unit price + freight (Class 9 dangerous goods handling is not optional and not cheap) + UN-certified packaging + duty + insurance + the carrying cost of a minimum order quantity you didn't want + incoming inspection + whatever the exchange rate does between quote and payment.
Real example from our 2023 files: Vendor A quoted higher per kWh than Vendor B. Vendor B required three times the MOQ, charged extra for UN packaging, and shipped from a port with two extra DG handling fees. On paper B was 6% cheaper per unit. Landed, it was about 9% more expensive, and we'd have been sitting on eleven months of inventory. That gap was entirely in the fine print.
Step 6: Look at the pack build, not just the cells
Cell quality gets all the attention. Assembly is where I've had more failures. Check welding method (laser vs spot, and the pull-test data), nickel strip or busbar cross-section, harness routing and strain relief, compression on prismatic cells, thermal interface material, terminal torque spec, gasket and IP rating, and — critically — the BMS brand plus whether you can get firmware access and a protocol document.
A good BMS you can't talk to is a black box with a warranty attached.
Step 7: Audit the factory, or pay someone to
This is the step most buyers skip, and it's the one I'd protect hardest. You want to see the cell grading process, the aging and test room with real data logs, ESD controls, and batch traceability from cell code to finished pack.
Ask specifically how they define grade A and what happens to the cells that don't make it. If nobody can answer, the answer is that they get sold anyway.
Step 8: Put warranty, failure process, and change control in writing
Three clauses matter more than the headline warranty length. First, capacity retention with a defined test method — for example, 80% of rated capacity at 3 years measured at 0.5C and 25°C. Second, who pays freight on a return, and how fast. Third, a no-silent-substitution clause: cell model, BMS firmware version, and manufacturing site cannot change without your written approval.
The vendor is "flexible." What I mean is they'll re-quote and swap components once you've stopped paying attention. Change control is the clause that stops it.
Step 9: Pilot order before volume
Take 5–10% of your annual volume, from the same factory, against the same signed spec. Inspect it. Measure it on arrival with the same test you ran on the samples, and compare the numbers side by side. If the pilot and the sample don't match, you've found out while the order was still small.
Five mistakes I've made or watched other buyers make
- Accepting "grade A" without a definition. It's a marketing term until someone shows you the grading criteria and the test data.
- Using a pack-level cycle life number. The pack is only as good as the weakest cell in the weakest module.
- Forgetting the DG freight. Many forwarders won't touch loose cells at all. Confirm your lane before you confirm your price.
- No incoming inspection. A basic capacity tester costs less than one bad delivery, and it's the only data you own.
- Quoting from an old price sheet. Cell pricing has moved enough in the last two years that a 2023 quote is basically fiction. Get three fresh quotes, date them, and note that prices move.
Looking back, I should have paid for third-party inspection on our very first container order. At the time, the vendor's own QC report looked thorough and the samples were fine — which is exactly the problem with samples from a plant you haven't seen. We spent about $400 on the inspection later and recovered roughly that much in one corrected shipment. Cheap lesson, learned late.
Honestly, I'm not sure why some vendors consistently hit their cycle-life claims and others hedge every number. My best guess is it comes down to how disciplined their cell grading is, but I've never gotten a straight answer and I've stopped expecting one. What I do instead is test, document, and make the spec contractual.
None of this is clever. It's just the paperwork that keeps a cheap quote from turning into an expensive year.