I work at a medical device company. We use no analytical test methods, we build electro mechanical devices. However, we have been challenged by FDA on this topic. My experience with them is that the expectation is that we have proof that our "test methods/test procedures/measurements" produce reliable answers. We have taken this to mean the test procedures
(test methods) we use to collect data, whether a measurement as part of a design verification test, a measurement used in process validation, or a measurement performed on the product at final QC, need to produce an answer that is reliable.
In my opinion, most of our measurements are direct (physical dimensions, tensile force, electrical resistance) and can be performed with general lab equipment designed for that purpose. These are calibrated so the accuracy, precision, bias... sections of TMV are taken care of. When we are using them in a way they were not intended - like using the resistance of a fluid to determine the purinty of fluid - the calibrated instrument may be accurate, but where is the proof that the purity of the fluid is being reliably determined? This is what TMV would provide.
Now, if we are using a piece of test equipment that we designed in house, we must now verify all the elements of TMV.
One other element we review for apparent straightforward measurements is if it is easy for the person performing the measurement to influence the result. Example, the operator's ability to use an optical comparator for precise measurements in a room with no ability to reduce the ambient light - how consistently can he repeat his measurements, and can they be be replicated by another person using the same optical comparator? Gage R&R (a part of TMV) can take care of this.
Because to the lack of guidance we have endless discussions on this topic; Quality Engineering leans toward TMV on every measurement, Product Engineering leans towards only doing a TMV on custom measurement equipment.
From consultants I have spoken to on this subject - be able to defend you test measurement results. That is, if you did not perform a TMV - why not?