I set up predictive maintenance systems on a regular basis and you have arrowed in on the biggest error that most people make when they set these systems up.
Before you pick up an analyser or any sensor you should make an assessment of which failures are going to happen, identifying the symptoms of those failures and only then should you start deciding which readings to take.
On a standard induction motor there are a limited number of failures that can happen. Of those failures, some are more likely to happen in specific operational scenarios. For example - you are more likely to see broken rotor bars on motors with frequent starts.
Once you have identified which failures are likely to affect your motor THEN you decide which readings to take. I appreciate the several comments posted above and they are all valid. My point is that you will get the best "bang for your buck" if you use your head.
I presented POTENTIAL FAILURE ANALYSIS at the Entek National User Group conference (remember Entek) in Cincinnati in 1997. At that time most people saw this as some form of Rocket Science but we have seen continual improvements in the systems we have set up because of this procedure.
Good luck and just remember to use the posts above sensibly.