Six more.
The easy answer is wrong.

Shaft currents. Strip the bearing and the race is covered in fine grey fluting, a washboard of evenly spaced pits across the track, not the bright spall of a mechanical failure. The inverter puts a voltage on the shaft and it discharges through the bearing to earth, sparking through the grease thousands of times a second until the steel erodes. A new bearing just restarts the clock.

The tell: evenly spaced frosted fluting and grey grease, not one bright crater.

Fix: a shaft grounding ring or an insulated bearing, plus a screened cable glanded properly. Then it lasts.

Build up. There's a hard crust of caked product on one side of the tail pulley, so that side sits at a bigger diameter and the belt climbs off it like it's running on a cone. Every tracking tweak fights the lump instead of removing it. Scrape the pulley clean and the belt centres itself.

The tell: it always walks the same way, and the tail pulley is crowned with hardened muck on one side.

Fix: clean the pulley, fit a scraper or plough, stop chasing it with the tracking bolts.

Wrong sensor for the mount. It's a non flush (unshielded) prox buried flush in a steel block, so the surrounding metal saturates the field and robs it of range. It only reads when everything's cold and the target's dead centre. Fit a flush rated sensor, or sink the mount so the face stands proud, and it reads every time.

The tell: it works on the bench and fails in the fixture, and the face sits level with the metal around it.

Fix: match the sensor to the mount, flush or non flush. Nothing wrong with the sensor.

Imbalance. The impeller is caked with a thick, uneven layer of overspray and dust, heavier on some blades than others, so it's spinning like a wheel with a weight welded to the rim. That force went straight into the bearings that got the blame, and now it's into the welds. New bearings on an unbalanced wheel is just a countdown.

The tell: vibration that crept up over weeks and blades with an uneven crust, not a sudden bearing rumble.

Fix: strip and clean the impeller, then check balance. Bearings only after the wheel runs true.

No float. On the rebuild both bearings were set as fixed instead of leaving one as a floating expansion bearing. When the shaft heats up and grows there's nowhere for it to go, so it loads the bearings hard along the shaft until they overheat and fail. The bearings are fine, the setup traps them.

The tell: both ends hot and failing together after a rebuild, on a shaft that sees heat.

Fix: free off one housing as the float side, locate only the other. Let the shaft grow.

Air quality. Crack the valve open and the spool and seats are gummed with oily sludge and rust carried down the line. There's no proper filter regulator ahead of it and the compressor is passing water and oil, so every new valve is fouled the moment it's fitted. The valve isn't failing, the air is filthy.

The tell: sticky brown residue on the spool and water in the airline, same fault across more than one valve.

Fix: fit and drain a filter regulator, sort the water at the compressor, then the valve lasts.