mike_nyc said:

The pattern that distinguishes a bad batch from an exit is behaviour rather than product.

This is where I part company with the consensus forming above. Import rules are jurisdiction-specific and this board keeps giving US-shaped answers to non-US questions. What is a personal-import allowance in one country is a controlled-import offence in another.

Worth separating that from vendor vetting, which this thread keeps folding into the same question. They behave differently and the advice does not transfer.

I would rather be corrected than agreed with, if it comes to it.

One concrete data point for the thread. Red flags, in rough order of how much they should worry you: no verifiable address, no batch numbers, prices far under market, vendor-commissioned tests only, pressure toward irreversible payment, and shipping with no temperature control.

julia.endo said:

Import rules are jurisdiction-specific and this board keeps giving US-shaped answers to non-US questions.

Adding the part of the answer the thread has not reached. The useful checklist is about verifiability rather than presentation. Independent test results commissioned by buyers rather than by the seller, consistency across multiple batches over time rather than one flattering report, a physical address and a company registration that resolve, batch and lot numbers on the actual labelling, and a shipping practice that matches what the material needs. Everything a good-looking website provides is cheap to fake; none of the items above are.

Worth separating that from vendor vetting, which this thread keeps folding into the same question. They behave differently and the advice does not transfer.

A narrower follow-up, since the general answer is now clear:

What a vetting checklist should actually contain, as opposed to a list of things that are easy to fake?

Closing the loop on my own question.

Went back through my own correspondence with an earlier supplier and the warning signs were all there in the replies, weeks before the quality changed. I was reading the product and not the behaviour.