Names that arrive readable
Cyrillic, Chinese and accented European spellings turn to mush when the character set is guessed wrong, and nobody notices until the contacts are already on the handset. Detection starts at byte order mark, validates UTF-8, then weighs byte statistics for windows-1251, windows-1252 and GBK. Override the guess yourself if it looks off. vCard 2.1 output also gets CHARSET=UTF-8;ENCODING=QUOTED-PRINTABLE, so an old handset shows the letter itself and not =C3=BC.