Accepted answer
It helps to be literal here: what a useful certificate carries is debatable, but there are certain fields that almost no research-grade certificate includes despite their being standard in pharmaceutical work.
Lot number visibility is the first check, and it is the one most certificates fail — the number is not printed on the vial, or it is on the certificate but the vial is unmarked.
Multiple testing on the same lot reported in a single document suggests internal quality control; identical figures across different lots suggests the same certificate reused.
Published inter-laboratory comparisons of the same material routinely find spreads of half a per cent to a full per cent, which sets realistic expectations.
Be careful not to over-read this — a certificate that is imperfect is not worthless, it is just a starting point for thought rather than an endpoint.
My working rule is that a document without a lot number, a method section and a test date is marketing.
edited 10 Nov 2024 by h_villanueva — added a caveat about sampling