Yeah i was about to post about what harry said. The headline figure is more or less ... "80% of our members have been sober for at least a year", but the caveat is that only 5% of people who walk through the door last a year. So ... only something like 4% of people who walk through the door seeking help actually stay sober for a year.
But that doesn't stop them variously claiming the program is 80% successful, or even 100% successful. e.g. they hand-wave all people who aren't helped by labelling the failures as "unsavable". i.e. the program doesn't fail you, you fail the program, and they say "100% of people who stick with the program stay sober". So the program works 100% of the time for those people for whom the program worked. And if you didn't stick with the program, it's your fault, not the program which is "perfect wisdom from the holy founders of AA" and you're unsavable filth headed for hell, and other AA members are encouraged to sever all personal ties with you!
Also, any controlled trial to date has shown that sending people to AA has no more chance of leading to sober outcomes than a control group. e.g. (not the exact figures) 5% of both groups will quit drinking, but the ones who went to AA ascribe this purely-by-chance occurance as being due to AA. So the whole thing is about them tricking you with self-selection bias that they helped you.
Actually in a huge meta-study of 48 known alcohol addiction treatments, AA came 38th, just behind "attention placebo". i.e. give someone any fucking thing to do as a hobby, and it's about as effective as AA at preventing them relapsing into drinking. Actually more so, since a hobby picked completely at random gives them less time socializing with fellow drinkers, unlike AA. In AA all your time ends up being around fellow humans with drinking problems, and they actually encourage you to cut off your non-AA friends. So you end up in a huge circle of addicts. This is basically the opposite of helping.