I almost quit at week five.
I'd convinced myself it wasn't doing anything. Then someone explained that the research ran twelve straight weeks before they measured a single thing — and that I was judging it less than halfway in. I stayed. I'm glad I gave it the time it actually asks for.