One thing is that this is not really 6-12 pregnancies out of 100 (possible pregnancies), it's 6-12 women out of 100 using the pill in a given year will get pregnant in that year, so it's pretty likely that even if the women you know had a similar distribution of use patterns as the average in the studies, there's a decent chance none of them would get pregnant. If you know 40 women and assume a 5% failure rate, there's still a (19/20)^40 = 13% chance that
none of them would get pregnant.
Add to that that the people you know might be atypical in their use patterns. The "perfect use" failure rate for the pill is ~0.3%, which I take to mean that most of the pregnancies that occur on the pill are because someone forgot to take pills, that sort of thing.