IMO, 'fail' is a bit loaded in this context. The ability for even a small minority to fork off out of a perceived existential necessity is absolutely vital to the health of bitcoin overall. (though, I'm not convinced such a necessity is actually being perceived by most bip110 supporters)
That being said, if bip110 were to split off with say 5% of the hash rate, I find it hard to imagine them being satisfied with the corresponding 200-minute block times until they finally reached their first difficulty adjustment ~40 weeks later. I think it's much more likely in such a scenario that they would just pivot to a hard fork instead in order to adjust the difficulty right off the bat (which would then fairly be viewed as a failed soft fork attempt).
