There is a table in the book Handbook of Alcoholism Treatment Approaches comparing various treatments of Alcoholism; it ranks AA as being one more the most ineffective treatments.
This table is, shall we say, weird. It rates acupuncture as one of the top 20 treatments of alcoholism, and gives motivational enhancement (#2 in the table), cognitive therapy (#13), and AA (#37) very different ratings, when the 2006 Cochrane study rate those as being about equal; other studies rate AA and/or TSF (Twelve Step Facilitation: teaching patients how to go to meetings and be a part of the AA culture) as being superior.
I also note that they only looked at seven different studies when rating AA, which leads me to believe they have only looked at old randomized controlled trials which do a terrible job of measuring AA's effectiveness.
I mean, if there was something besides this book out there arguing that acupuncture was a good deal more effective than AA, than maybe this table would convince me. But the results seem to be randomly placed on the table and I do not know of any other meta-study ranking AA as being inferior to other treatment methods. On the other hand, Cochrane 2006 said that AA is about as effective as other treatments, and Project MATCH shows that TSF is somewhat more effective than other therapies (See PMC2746426 for discussion). Fiorentine 1999, Vaillant 1995, and, yes AA's own big book give a success rate for AA around 75% for people who choose to attend one or more meeting a week; Moos & Moos 2006, which measures something slightly different, has similar figures.
Slate Star Codex discusses this table in depth in section VI (scroll down).