Most self-help, if measured by titles, is probably terrible simply by Sturgeon's Law. But is most self-help, as measured by sales? I haven't looked at sales figures, but I imagine it's not that unlikely that half of all self-help books actually consumed are the ones that are genuinely helpful.

Another complication is that Sturgeon's Law applies as much to the readers. The dropout rate on free MOOCs is astronomical. (Gated link, may not be accessible to all.) "When the first Mooc came out, 100,000 people signed up but “not even half went to the first lecture, let alone completed all the lectures.” "Only 4-5 per cent of the people who sign up for a course at Coursera ,,, get to the end."

Picking up a self-help book is as easy as signing up for a MOOC. How many buyers read even the first chapter, let alone get to the end, and do all the work on the way?