I think you are misunderstanding me... you claim to have "perfected" your drivetrain losses, without test data. Since the differences are too small to be discerned by the rider, how could you know this is the case? Spinning parts in your hand certainly won't do it... it has to be under load.
That does not mean there is no test data. Data taken from one set up can be ported to the next one and so on.
It isn't exactly rocket science either.
If a new part spins very easily, it probably has crappy seals and light grease... and will need to be overhauled more frequently, while providing very little resistance benefit.
Probably always the case with industrial bearings: when it spins suspiciously easy it more often than not tells you it is of the low tolerance, low quality metal retainer type.
The problem with these industrial bearings is that no matter how good they are, none of them are designed with our kind of use in mind.
You occasionally come across some that have been made to order or carefully selected for our kind of application but I bet that most guys in the industry just can't be bothered.
Which is why these measurements are so important. If it clearly shows that better bearings/lubricants/seals etc. are wortwhile we'll at long last have some leverage.
@Sawyer: You can safely say that there is no correlation.