The short version: same compound, same laboratory, same method, ideally same week — otherwise you are comparing laboratories rather than suppliers.
Content is the more discriminating measurement than purity for supplier comparison, because purity clusters tightly among competent suppliers and content does not.
Lead time and lane behaviour are supplier properties too and are easier to compare than analytical ones, because they need no laboratory at all.
Comparisons drawn from different laboratories at different times are weaker evidence than most people treat them as.
Use a fixed documentation checklist rather than an impression.