I've updated customers in yC4C using the massdata maintenance tool.
1st: Export of selected data: The data contained strange symbols.
--> 'ü' was transformed to 'Ã¼'
--> 'ß' was transformed to 'ÃŸ'
2nd: Import of updated and corrected (mutated vowels) data to yC4C.
Result: For names containing any mutated vowels those were replaced by '#'!
How to avoid such an issue for the future. Due to that circumstance we have a lot of manual effort to identify and correct every single dataset.