I am not sure I have a definitive position on this, however I believe the following:
Mainstream adoption will not occur until dapps can provide utility beyond that of normal apps, to the point that people actually prefer to use dapps over apps.
We could have a 'killer dapp' developed that overnight results in mainstream adoption.
I know that that sounds like a bit of a hedge but I think those are the two possible outcomes.
Oh I don't think you're far off at all. Number 1 is a given, first we'll need a lot of useful dapps that outperform the current apps, which is what a lot of people are trying to build.
Like what I wrote in the post that only a small portion of that 84% have actually gone live, the rest is still in development, testing, and research... But when the big wave of innovation arrive, adoption would be much closer.
Thanks for the awesome comment btw