Two Thursdays ago, a voice note came in on WhatsApp from Elena, who runs an eight-bay bodyshop in Antwerp and has been buying Chinese EV panels from us since 2021. Her message was short: “Forty percent of the panels that arrived last month had transit damage or paint faults. My supplier tells me their return rate is under 2%. Either they are counting something else, or they are not counting at all. I cannot tell which, and I have no way to find out.”
Elena had done nothing wrong. She asked for a return rate before the order, got a number, and treated it as a fact. The trouble is that a supplier’s return rate and a buyer’s return rate are two different measurements of the same shipment, and the gap between them is where the money disappears.
So this guide is about closing that gap: what a realistic return rate looks like at each level of the supply chain, how to measure your own number instead of trusting somebody else’s, and what to write into the purchase order so that a dispute, if it comes, has a figure attached to it rather than an argument.
Everything below comes from our own shipping records going back to 2013 — not from published industry statistics, which for this corner of the trade are largely nonexistent. The numbers are ranges, because a range is what a real ledger gives you.
The Four Tiers, and What Each One Costs You When Something Arrives Broken
There are effectively four ways to buy EV parts from China. They get described by the same phrase — “factory direct” — and they behave nothing alike once a part arrives damaged.
Tier 1 — Factory direct, your name on the purchase order
Your company name is on the purchase order, the invoice comes from the factory, and the factory’s own people pack the crate. In our ledger this tier returns the least, because there is nobody between you and the people who physically inspect the parts. Our realistic range is 0.5–2% claimed units per 100 units received. Below that, in my experience, something is wrong with the measurement, not with the parts.
This tier is the rarest one people think they are in. If your invoice says “factory” but the packing list carries a trading company’s name, you are in Tier 2, not Tier 1.
Tier 2 — Factory-direct sourcing through a trading company that owns the goods
This is the most common structure in the market and the one most often mislabelled. The trading company buys from the factory, holds the title, and sells to you. The parts are usually genuine factory output, but the last inspection happened before the goods entered a warehouse, and from that warehouse they travel through a freight forwarder and sometimes a second consolidator. Our range here is 2–4%, and almost all of the damage is logistical rather than dimensional: crushed corners, water ingress, scuffed paint on a panel that was stacked without separators.
The AVATR 11 full body panel set is a good illustration of why. Four doors, hood and tailgate travel together, which means the hood is resting on something and the tailgate is carrying the compression. Paint protection on the edges costs very little at the factory and almost nothing at the trading company. If the seller will not say who packed the crate, ask.
Tier 3 — Trading company reselling without a factory relationship
Here the supplier is a business buying from other trading companies, sometimes on terms, sometimes assembling mixed lots. You are getting whatever the market gave them. Expect 4–8%, and expect the defects to be more varied: wrong revision, superseded part number, missing hardware, a component that is genuine but fitted to a different model year. This is the tier where fitment discipline pays for itself.
The Toyota bZ4X full body panel set is a part number I would never accept on a verbal confirmation. The bZ4X range carries multiple revisions across model years, and a panel set can be physically correct for the car and still be the wrong revision for the build you are working on. A number plus a fitment note is not bureaucracy; it is the cheapest insurance in the transaction.
Tier 4 — Marketplace, consignment and salvaged stock
Consignment stock from a vehicle dismantling operation, auction lots, and marketplace resellers. The honest range is 8–20%, and for panels it goes higher, because you are buying a part that has already been removed from a car. Fits, but carries history: previous repair paint, a misaligned aperture, a sensor mount that has been twisted. Buy it when the part is genuinely unavailable and the fitment is trivial — a door, a hood, a bumper cover. Do not buy it for a structural or high-voltage-adjacent item.
I would include the Deepal S05 full body panel set in a different conversation: a model launched recently, with a short and thin supply of used parts, where the only sensible source for most buyers is new stock from a factory relationship. When a model is this new, the aftermarket tier structure has not formed yet — you are choosing between a factory channel and gambling.
What the Damage Actually Costs You by Category
Tier tells you the quality of the channel. Category tells you where the loss concentrates. Across our shipments the claim rate is not evenly distributed at all, and a buyer who averages across their whole order will misjudge both their supplier and their own stock.
| Category | Realistic claim rate | Dominant failure mode |
|---|---|---|
| Body panels (doors, hood, tailgate) | 4–9% | Transit damage, paint fault from stacking without separators |
| Glazing and quarter windows | 2–4% | Breakage in transit, carrier or channel supplied separately |
| Harnesses and cable sets | 3–7% | Pin errors at the connector, wrong length branch |
| Modules and control units | 1–3% | Firmware revision or pairing failure on install |
| Mechanical running gear | 0.5–2% | Fitment variant, not manufacturing |
Notice that the two highest-rate categories — panels and glazing — are also the two where a claim is expensive to argue about, because the photograph has to show the damage and the carton label in the same frame for the supplier to accept it.
Mechanical parts are the opposite. A rear brake disc for the Toyota bZ4X that is the wrong revision comes back cheaply and quickly, because a serial number on the disc settles the argument. That is why we push customers towards numbered mechanical items rather than towards panels, and why a good buyer negotiates harder on the small high-volume SKUs first.
The 90-Day Method for Measuring Your Own Return Rate
You cannot audit a supplier’s claim rate. You can measure your own, and if you measure it properly you can infer theirs. Here is the method we ask customers to run, and the one we run ourselves.
Step 1 — Count what arrives, not what was ordered. Open every carton on the day it arrives. Log the quantity received per SKU against the purchase order, on the arrival date. Quantities that quietly differ from the invoice are often the first signal of a short-ship or a double-ship, and they change your denominator.
Step 2 — Log every defect on arrival, before the part is fitted. Use a fixed taxonomy, because the argument later is always about categorisation: transit damage, paint fault, fitment mismatch, missing hardware, electrical fault, firmware or pairing failure, wrong part shipped. Photos at the moment of opening, carton label visible in frame for anything you intend to claim.
Step 3 — Count claims filed, not replacements sent. This is where most buyer numbers collapse. If your supplier replaces a damaged panel without a claim being logged, your ledger records a good delivery. A claim has to exist as a document with a date and a reference before it counts.
Step 4 — Keep the denominator honest. Claim rate = units claimed, divided by units received, for that category, over a 30-day window. Report it per 100 units. If the volume in a window is below 20 units, say so and do not act on the number — one bad crate moves a 5-unit sample by 20 percentage points.
Step 5 — Compare, then ask for their ledger. If your measured rate is more than three times the rate your supplier quotes, that is a conversation. Ask to see their claim log for the same period and the same SKUs. In thirteen years I have never seen a supplier refuse, and I have rarely seen a supplier present a ledger that matches the quoted figure.
Step 6 — Hold the payment point until the window closes. The practical reason buyers never measure properly is that the money has already left. If you settle at delivery, the 30-day claim window is a formality. If a portion of the balance settles 45 days after arrival, the claim window is a live instrument.
Six Terms That Decide Who Pays When a Part Is Wrong
Whatever you measure, the measurement only pays out if the purchase order says so. These six clauses cover almost every dispute I have seen on this site’s behalf.
1. Define the claim window by arrival, not invoice. “Claims accepted within 30 days of arrival at the named port or warehouse.” Invoice dates and arrival dates can be weeks apart, and the gap is where time runs out.
2. Specify the evidence. Photographs showing the packaging, the carton label, and the defect in the same frame, plus the serial or batch number where the part carries one. A supplier who says “send us photos” without specifying this will keep rejecting photos, and you will conclude that claims are not worth filing.
3. Split the freight explicitly. Who pays the return shipment, and who pays the replacement shipment. In my experience the single most common silent disagreement is not about the part at all, but about the freight, and it is the fastest way to turn a 3% claim rate into a 0% claim rate: the buyer quietly stops filing.
4. State the remedy order. Replacement first, then credit, then repair, and give a deadline for each. A supplier who has to choose between an RMA process and a credit note will choose the one that costs more, and you will pay for that choice in the price.
5. Require batch traceability. The supplier keeps the mapping from batch or serial numbers to your purchase orders for at least twelve months. Without it, a claim about a single panel from a 400-panel crate is unarguable in principle and unprovable in practice.
6. Add a threshold clause. Above a defined claim rate on defined SKUs, you get the right to withhold the balance pending a joint review, or to require third-party inspection at the supplier’s cost. Set the threshold in advance, in the contract, while the relationship is healthy.
Five Numbers That Should Make You Slow Down
A supplier claiming a return rate below 1% across a year and several hundred shipments is not describing a perfect operation. They are describing a buyer who stopped filing. A rate of exactly zero on mixed body panels over six months means the panels are not being opened on arrival.
A supplier who quotes a single blended rate for the whole catalogue is hiding the structure. Panels and brake discs do not have the same rate, and a blended number tells you nothing about where your money went.
A supplier who asks you to pay the return freight before evaluation is shifting the cost onto the only party with an incentive to stop claiming. That clause should have ended the conversation before it started.
A supplier who cannot produce pre-shipment photographs of the actual crates, with the actual labels visible, is asking you to accept a process they cannot evidence. We send that set with every panel order we ship, and it removes more disputes than any contract clause I have written.
And a supplier whose fitment notes are one word — “universal”, “compatible”, “same” — is telling you the claim rate is unknown. Fitment problems do not arrive as claims; they arrive as a part on your bench that will not go on the car. One front brake tube for the Toyota bZ4X with a routing that does not match the chassis is a bench day lost, and it never appears in a return figure because it was never returned.
What We Do Differently, and Why It Moves the Number
Our own claim rate against Tier 2 suppliers has fallen by roughly two thirds over six years, and none of it came from negotiating harder. It came from three operational changes that any buyer can copy.
Pre-shipment photo sets. Every panel order is photographed after packing, before the crate is sealed, with the crate label in frame. When a claim arrives in Antwerp or Rotterdam we compare it against that set before anything else. Half of our transit claims are resolved by the photograph alone, in under an hour, without the supplier needing to accept anything.
Batch numbers on the packing list, not just the invoice. A crate of forty panels arrives with the batch mapping printed on the packing list. This is what makes a claim against a single panel in a large consignment provable rather than arguable, and it is why our dispute rate with repeat customers sits well below the industry pattern.
Fitment stated in plain language on the line. Not “fits bZ4X” but “fits bZ4X, 2023–2025 build, excludes the sport trim”. It takes us longer to write and it removes a whole category of disputes that no return rate can capture, because the part never leaves the shelf. A steering wheel assembly for the Toyota bZ4X carries at least three distinct codes across the range, and the difference between them is trim and build year, not the part name.
The same logic applies to electrical items, where the failure mode is never dimensional. A chassis harness for the ZEEKR 7X can be entirely correct and still fail on installation because a pin at one connector was populated differently on that batch. A ZEEKR 009 electric compressor harness behaves the same way. These parts rarely generate claims, because most faults are diagnosed and fixed at install — but they generate your worst days, and no return rate will ever show them.
And where the part is a high-voltage charging module, the pairing question dominates everything else. An on-board charger for the Li Auto L7 and a high and low voltage charging system for the ZEEKR 001 can both be genuine, both be the right revision, and still be refused by the vehicle because the software pairing step was skipped. That is not a supplier failure and not a return, but from the workshop’s side it looks exactly like one, which is why a reliable supplier tells you about this before you install the part rather than after you have the diagnostic tool connected.
If you are running a workshop or a small distribution business and you want to know where your own number really sits, send me your last three shipments with the quantity received and the quantity claimed. I will tell you what the rate looks like against a Tier 1 and a Tier 2 baseline for those categories — and if the answer is that your supplier is simply not being asked the right questions, I will say that too.
Because the cheapest part in the world is not the one with the lowest invoice price. It is the one that arrives once, fits the first time, and never comes back to your desk at all.
— Wang (JLL Auto Parts founder) · 13 years sourcing EV parts from Shanghai · jllautoparts.com

