The article itself doesn't use the phrase "conversion rate" or even the word "conversion". Referring to the 21% as the conversion rate happened right in this thread.
I have no trouble believing that 21% of Pinterest users have indeed purchased something that they saw on someone's Pinboard ... but remember that they likely viewed a great many Pinboards and visited a great many sites along the way before finally making that purchase.
The numbers are "wa-a-ay far apart", but the problem I see is not the numbers themselves, but sloppy naming. It's just not accurate, IMHO, to call that 21% a "conversion rate".