IIRC, when the user disables/deletes his/hers OkC account, they can specify the reason: "met someone on OkC (they can type the other person username)", "met someone elsewhere", "not interested anymore", etc. So they probably have lots of good data on success.
Self-reporting in general is fraught with problems, doubly (quadruply) so when it's something as close to the ego as dating.
Also assuming the account is shut off at all as opposed to simply idled.
Real success is extremely hard to quantify, which is why i'm not suggesting that as the goal. The goal is to grade spam messages, and encourage people to not repeat known spam patterns. (And, if you show the score to the recipients, it also acts as a spam filter.)
Encouraging / forcing people to write less spammy messages might improve the overall message quality, just by increasing the cost, in effort and time, of sending the messages. It also provides some minimum level of feedback to the people who get zero responses from their messages.
lol