How 'bout 'Control of Nonconforming Tests'? I assume you are referring to a test performed incorrectly as opposed to test failures. If an item 'fails' a test but the test was performed correctly, there is no nonconformance on the part of the testing lab. Any failure of the test item is an issue for the customer to deal with.
Items should be well identified but how far you go, especially with respect to evidence (pictures, data, etc.), depends upon the overall requirements. The item its self is 'evidence' and it is the customer's property. Typically a lab will ask ahead of time if the customer wants the remnants of the test. Typically they do, especially if the DUT (device under test) fails so they can analyze the failure mode(s). Just like any contract review process, all the customer requirements should be determined well in advance of the actual testing.
I once managed Cincinnati Electronics' environmental test lab - vibration, shock, Mil-S-901 Shock, humidity, altitude, explosive atmosphere, salt spray, thermal shock - you name it we did it. Mostly Mil testing (such as various Mil-Std-810 testing) but we took outside commercial work. As an 'Oh My' fact, I first started in automotive there, strange as it seems today, doing studies for Ford in environmental stress screening of a number of their electronic assemblies.
We also did the studies for qualification of the original Ryan Storm-Scope for aircraft. We took pictures of both the display and the DUT periodically throughout the testing process, took notes, we took electical readings. We ended up with a pile of documentation which we submitted with a long 'report' explaining exactly what we did, how we did it, when we did it, etc. The report was to be part of their submission to the FAA for 'qualification'.
The requirements were all determined prior to my replying to a RFQ. It was in that job that I first understood the importance of contract review.