Additional examples from when I was mechanical lab supervisor in a stainless steel mill - we were accredited by A2LA to ISO Guide 25 and our scope included tests for which there were no national or consensus standards. Among the tests without national standards were a "ridging test" a test developed internally within the company that dealt with formability of stainless sheet - it was a visual comparison test and our visual comparison standards were developed and controlled internally. Results also were reported externally to some customers. We also tested samples to determine a "modul-r" value that was reported to external customers (Modul-r was accepted by many customers in place of a strech-r value, which has an ASTM test method. Modul-r was actually more repeatable, quicker, and non-destuctive of test specimens than running a stretch-r, which involves preparing and pulling 3 tensile specimens ( 1 paralell to, 1 perpendicular to and 1 at 45 degrees to the rolling direction for flat sheet) and then making calculations to determine stretch-r) Again a formability test that was I believe originally developed by USX, By the time I took the lab over Tinius-Olsen had purchased the rights to manufacture the testers and associated punches, and ended up developing and certifying a set of standards, but as there were no national test methods based on this test method there were no national standards.
I also believe some standards for spectrometers, etc. have been developed by consensus testing between labs.