I am guessing the special characters are smart quotes and similar stuff. Those will definitely be lossy if converted to ASCII from UTF-8. He could translate to CP 1252, which has those characters in a non-standard location. The characters probably came from that encoding in Microsoft products.

Alternatively, he could replace the characters with ASCII equivalents. Smart quotes with normal quotes, en-dashes with hyphens, etc. This would be similar to the conversion that people do with the CP 1252 but using Unicode characters instead.