Check the batch reference matches
The first check is whether the certificate refers to the material actually delivered. Certificates covering a different batch, or a generic typical-values sheet presented as a certificate, are more common than they should be. Matching the reference against the delivery documentation takes seconds and catches a fundamental error.
Look at where results sit, not just whether they pass
A result at the very edge of a specification limit carries different information from one comfortably central, even though both pass. A batch consistently hugging one limit indicates a process running close to its capability. Reading position rather than pass/fail is what turns certificates into a monitoring tool.
Confirm the methods are the agreed ones
Results are only comparable when generated by the same method. Where a certificate names a different method from the one agreed, or names none, the numbers cannot be compared with your own or with previous deliveries. This is worth resolving at qualification so that every subsequent certificate is directly usable.
Check the date and the testing laboratory
Testing performed long after production, or by a laboratory other than the one agreed, both warrant a question. Neither necessarily indicates a problem, but both change what the certificate evidences. Recording who tested and when allows a pattern to be spotted if the arrangement drifts.
Watch for properties quietly dropped
Certificates evolve, and a property that appeared on early deliveries can disappear from later ones without announcement. Comparing a new certificate against an earlier one, rather than reading it in isolation, catches omissions that a specification check alone will not because the specification is silent on what must be reported.
Trend the values across deliveries
The single most useful thing to do with certificates is to keep the numbers and plot them. Drift, widening spread and step changes all appear in a trend long before any individual result fails. That early warning is the difference between a scheduled conversation and an unplanned investigation.
The single most useful thing to do with certificates is to keep the numbers and plot them. Drift, widening spread and step changes all appear in a trend long before any individual result fails. That early warning is the difference between a scheduled conversation and an unplanned investigation.
