AA may be the most effective of the options studied but that doesn't mean it is a successful program. It just means it is the most effective of what are generally non-successful programs.
Chronic Alcoholics had limbs (hands) cut off during early history. It wasn't very successful comparatively or maybe it wasn't "the best". The best known behavioral solution is the best behavioral solution qualitatively (which is a mix of programs, statistically). But calling such a program (as AA) something non-successful means you've using different criterion than the bar of sustained sobriety and reduced frequency of relapse. It's not clear what differentiates successful from non-successful in your estimation?