The useful measures are the ones that change before a failure, like vibration, temperature, pressure, run hours and the results of routine inspections. Tracking these over time lets you see when equipment is trending toward a problem rather than waiting for it to fail.
It also helps to track the basic maintenance measures, like how much maintenance is planned versus reactive and how often the same equipment fails. Those show whether the shift toward predictive maintenance is actually happening, and most teams find that consistent inspection data is the foundation they need before more advanced monitoring pays off.