If the stuff coming in from your web clients is using the "%XX" notation for utf8 character data, then any "wide" characters (requiring more than one byte in utf8) will require one "%XX" thingie per byte (e.g. a utf8 "ÿ" (U+00FF) would be "%C3%BF").

If you see that in your input, then pack("C",...) is the right thing as the first step: it creates the appropriate byte sequence for the intended utf8 character. The utf8::decode() step then handles the necessary step of getting perl to acknowledge that the given byte sequence should be treated as a utf8 character.