I looked through the literature and it seems that condoms are quite effective at preventing HSV infection when used, especially from male to female (on the order of 99%), and less so for female to male (around 60-70%). I would suspect the reason for failure is behavioral. People probably don't apply the condom until after most of foreplay is over and HSV-2 can be transmitted during "outercourse".
HSV-1 and -2 transmit through contact from skin location to skin. Including through asymptomatic shedding. Only sometimes is that location covered by a condom - in location and in timing.
See also: Wrestlers' "Herpes gladiatorum" which is just HSV-1. Which demonstrates how HSV-1 survives for transmission far more easily or broadly than just genital contact. Is there "HSV-2 gladiatorum"? Apparently it's usually HSV-1 but probably yes - neither HSV-1 or HSV-2 are really location specific.