I worked somewhere that accidentally ultrasonically washed a product batch with crystal oscillators in them, and the failure rate was probably even higher than 1%. It was very immediately noticeable when testing the product. There was also an above-average failure rate in the PMIC inductors.
After that, we moved the ultrasonic cleaner to a back room.
Just one XO per board is usually quite safe, but you do get those freaky highly sensitive outliers every once in a while. There are methods to reduce the risks.