I happen to have every field and each cutting of hay tested by a, National Forage Testing Association, certified Lab. To be sure I have not stood there and watched the process by the Lab but I understand after each 5 analysis a standard sample is analyzed and it must meet a deviation standard. If not the computer is recalibrated and the previous samples are re analyzed. The sample that the beam is looking at is dried to a known value and ground to pass a specified sieve size.
Not perfect but less costly and time consuming than one of the many Wet Analysis.
NIR or wet analysis the association submits unknown or blind samples on a regular basis to check the quality of the work.

In hay analysis Crude Protein is simply a measurement of nitrogen in the plant.

With, "Near Infrared Analysis Rolls Up its Sleeves and Goes Out to Work the Back 40" I presumed the handy little gadget is a NIR reading machine.
I suspect prices have come down for these neat tools. For one to sit in a laboratory or van used to cost more than $15,000.

I gather the dandy little tool is considerably more robust and stable than the computers and readers of 5 or 10 years ago. With out periodic calibration the changes would be duly noted and be of considerable interest, but not something to go to the bank with.

I did not mean to give the impression that tissue analysis is the one and only tool we need for managing soil fertility.
At every stop for a plant clipping a probe of dirt is also collected. Then the plant analysis and tissue analysis can be compared . Soil testing is good for determining pH, Organic Matter, Free Lime in the soil.
A soil test of 20 ppm P does not equate to 20 ppm P for a different soil type. The 20 ppm P on this soil will be reported as being different by other soil testing laboratories.
A soil test of 175 ppm K may be Very high in a sandy soil, but in many clay soils anything less than 350 ppm K is not just low but deficient.
As you may be alluding to, good records are very important. I consider records to be of equal to soil and plant analysis, combined.
A lot depends on the level of accuracy the farmer needs for his production and possibly marketing. Much of the hay going to dairies is very price sensitive to small variations in the feed value reported. When the seller has a report that his hay is 185 RFV and the lab used by the buyer reports a 175 RFV a heated discussion can develop.