A certificate without a method section is a claim rather than a measurement, and that distinction does most of the work here.
The pre-order sequence: request a lot-specific certificate for the exact line, check that it names a lot code rather than a batch family, check that it carries the column, gradient and detection wavelength, and check the analysis date is recent relative to the lot.
Ask what happens if a result comes back outside specification. The answer is informative regardless of what it is, and the question itself tells you something about how it lands.
The independent testing services this community references — Janoshik, Medutest, PeptideMeter and VendorInvestigate — publish results that can be searched by supplier, compound and date, which makes prior lots checkable.
The caveat is that a good result on one lot is a fact about that lot. It is a prior for the next one, not a guarantee.
Match the code on the glass to the code on the certificate. Everything else is secondary.