I think asking "what's the best way" is assuming it's just one thing. I see no reason it can't be the combination of a whole laundry list of things, just like "placebo effect" can be multiple things:

1. motivational effect ("I have an awesome new strategy! THIS CHANGES EVERYTHING")
2. regression to the mean ("I had an awful period, cast about for a new strategy, and simultaneously, things seem to be getting better! And of course correlation=causation")
3. survivorship bias (if 10 people start a useless thing simultaneously, after a few weeks a good chunk of them probably still think it's working; a few months later, there will be many fewer survivors...)
4. wishful thinking ("I had a bad few days - but they were exceptions and have a perfectly reasonable explanation X, Y, and Z, so I still have faith in this thing.")
5. time delay before meta-akrasia kicks in ("That part of me hasn't figured out the right excuses and tactics to defeat the new system, so it's working - temporarily.")

I'm sure you could suggest some more.