The problem with current techniques is that nothing works reliably. If you can go so high as to have a document that works to deconvert 10% of educated theists, then you can start examining for regularities in what worked and didn't work. The trouble is reaching that high initial bar.

It seems to me that Derren Brown once did some sort of demonstration in which he mass-converted some atheists to theists, and/or vice versa. Perhaps we should investigate what he did. ;-)