The industry as a whole is very "numbers phobic", in large part due to the potential issues they'd have in accurately marketing the small number of local matches typically exist among the active membership.
Case in point: pre-acquisition (by just a few days, really), OKCupid put out this scathing article about how useless it is to pay for a dating site, and drills down heavily into the numbers:
Why does my link not actually go to OKCupid? Well, Match.com bought them, and <poof>, the article disappeared!
Anyway, Long story short, I'd say your friend is right in terms of "active traffic" being in favor of Craigslist… After all, only 5% or less of match.com and eHarmony's respective members are likely active! That said, there's no way to tell how many of the posts on craigslist are real to begin with (the tales of scam/spam/cam ads are never-ending), as well as no way to judge things like number of CL marriages, etc. Oh yeah, and Craig also sells used dishwashers and rents apartments, among many many other things, so there's also no good way to separate out the dating traffic.
Perhaps it's easier to say that, assuming all active dating traffic on all dating sites was real, including craigslist, that the latter site would be far ahead of any of the pay sites in terms of worldwide ad postings.
Still, I've blatantly ignored the other majors, like free site plentyoffish.com, which is reputed to have a pretty large active membership, as well as sites like Zoosk and Tagged with fairly high "buzz" factor (well, at least, last year).
With no one releasing solid verifiable figures on their performance, it's just not really possible to say whether you or your friend are correct. Rather, it's best to say that there really are no actual "dating" sites – all the sites mentioned are simply compatibility matching engines, none of whom track or report statistics on actual dates that happen in the real world. Kinda sad, huh?