Our First Shipment of Solar PV Modules Cost Us 80,000

In November 2017, we shipped our first batch of solar PV modules — five hundred of them, rated 330 watts, polycrystalline — bound for Mombasa, Kenya, for a customer building distributed power stations in Nairobi.

I handled the shipment myself. We'd been selling used equipment; PV was a new category. The supplier was a second-tier factory in Jiangsu, quoting eight fen per watt below the first-tier makers. Five hundred modules, 460,000 yuan of goods.

We settled on the supplier in September. We'd looked at three before that: two quoted higher, one was too new. This factory had been running six years; the owner walked me through two production lines — clean workshop, workers who didn't chatter. I felt it was reliable at the time.

The inspection we skipped

Before shipment, the supplier sent the factory test reports — power, current, and voltage printed for every module, plus EL images. Beautifully made reports.

I supervised the loading myself that day — two containers, modules standing upright, wooden pallets wrapped with three passes of strapping. I felt quite settled at the time, figuring good packaging meant no problems. Looking back: packaging protects the corners, not the power output.

I asked the supplier whether third-party testing had been done. He said the factory self-tests every batch — trust the data — and that third-party testing needed scheduling: SGS had a twenty-day queue in November, which would miss the vessel.

The sailing was November 28. I did the math: waiting for third-party testing meant rebooking the vessel, and rebooking meant forfeiting the slot deposit. I said, ship it then.

The boss later asked me: did you look at the test report? I said yes. He asked: factory report or third-party report? I said the factory's. He said nothing. I still remember that look on his face.

That exchange ended up being worth 80,000 yuan.

Arrival testing came in 4.9% low

The goods reached Nairobi in late December. The customer's engineer sampled twenty modules and ran them on his own test bench. Results came back: average measured power 314 watts — 4.9% below nameplate. Not one of the twenty hit its rating.

The customer sent the test report, photos of the bench, and the calibration certificate, with one line in the email: per the contract, negative power deviation beyond three percent — compensate the price difference.

What does 4.9% mean? The customer bought a 165 kW plant at nameplate and actually got 157 kW. The grid-connection capacity changed, and his agreement with the power company had to change with it.

I had reviewed that contract myself — I wrote that clause in. I put it there to reassure the customer. I never imagined it would be used against us.

The supplier stonewalled

I went to the supplier. Their answer was three sentences: the test conditions were different — our report was done under STC; we don't know the customer's bench and don't accept it; microcracks may have occurred in transit, so responsibility doesn't necessarily lie with the factory.

I said fine — send retained samples from the same batch to a third party, 1,800 per module, 18,000 for ten; whoever is wrong pays. They refused, said it was unnecessary.

We held one video call. Their technical manager produced the calibration certificate for their own tester and claimed the customer's portable bench had large error. I had the customer's engineer photograph his calibration certificate: done by a German institution, calibrated six months prior, valid for a year. The supplier's side went silent.

Three weeks of wrangling, and the supplier finally conceded 20,000. Meanwhile the customer emailed daily — his schedule was locked, the grid-connection date signed in ink with the power company.

We carried the 80,000

On January 15, the boss made the call: we cover the difference. Converted at measured power, we compensated the customer 64,000 yuan, and we also covered his 16,000 in re-testing costs. Eighty thousand total.

Counting the supplier's 20,000, the customer recovered 104,000. Our gross margin on that order was 53,000. In other words, we lost 27,000 on the deal and threw in two months.

The day the money went through, the customer wrote back. He said he knew the supplier had only conceded 20,000 and the rest came from us. He said his next order would still come to us — on condition that goods go to a third party first.

I printed that email and pinned it to the wall facing my desk. Trust bought for 80,000 yuan is more solid than 80,000 yuan of advertising.

The rules afterward

From that shipment on, three things became mandatory before any PV module leaves.

One: randomly pull ten modules per container for third-party testing — power plus EL — and the report travels with the goods. Testing is 18,000 per container, under two fen per watt spread out.

Two: the contract spells out a list of testing institutions accepted by both sides. Nobody's own bench counts.

The first container under the new rules went out in March: ten modules to SGS, measured average power 0.3% above nameplate. The customer replied with two words: received. I read those two words several times.

Three: suppliers seal retained samples for six months. If a problem comes up, the seal is broken and samples go straight to testing; the losing side pays. This clause goes into the purchase contract — suppliers who won't sign it don't make our list.

The night he made the call, the boss said only one sentence: we wrote the clause, so we honor it. Honor it and the customer stays; refuse and this category dies.

That 80,000 was tuition for this set of rules. You only pay tuition once.