At some point in writing you can only plan so much before you have to write – it’s a matter of degree. This truth can frustrate some plotters, because you can only define so much before there’s nothing left to do. Your ideas may be totally wrong, your plan may be horrible, your plot awful – but you won’t know until you start writing.
Steven is echoing here something I've circled back to often: no plan for fiction survives first contact with the writing process. It doesn't mean plans are worthless. It only means plans are made in the present moment — typically, one where you haven't actually written a manuscript yet.
Something Steven notes: "I have trouble seeing how “pantsing” can work for complex stories, but perhaps I have something to learn there, no?" — This is worth commenting on separately, because it's indirectly related to how I've plotted several of my longer works.
One of my main rules about a work is, it doesn't matter how we get to the end (at least at first) as long as we get there. The main thing I have in mind with a story is what feeling it leaves you with while it's unfolding and once it's done, with the latter of those two being the most crucial. As long as we hit that ending, that's what matters.
My old analog of the road trip comes back to mind. We know that we're going to go west on Route 80 and eventually end up in California. We don't know the exact motel we're going to stop in during any given leg of the journey. We don't know which Pancake Hut we're going to eat in. We definitely don't know whether or not we're going to end up blowing a tire and needing a tow. But we know that we're headed west, that we want to end up in a certain place, and so on. A certain amount of improvisation towards that goal is needed and welcomed.
Now: once you have a general idea of what such a trip ended up being like, then you can drill down and refine and make sure all the little stops on the way complement the ultimate goal. I've gotten, I think, pretty good at sussing out whether or not a given detour or pit stop in a story serves its bigger needs, although sometimes I find whatever particular goal I had in mind early on in the process no longer applies later; snip! out it goes.
Now, Steve did specifically say complex stories. That could mean one of a number of things, not all of them what you might think. Complexity in a story is too often assumed to be convolution, as in a plot that is very knotty and full of double-reverses and whatnot. I tend to stay away from such things if only because I am not nearly smart enough to pull them off, but also because I have a different idea of what kind of complexity is relevant in a story. For me a story is complex if the pieces in it have a lot of thematic richness, or if the characters are multidimensional and humane. It's not if I need a map in the endpaper and a list of dramatis personae.
That kind of complexity, the kind that's about vertical depths rather than horizontal range, does lend itself to an improvisational approach. My tactic with this has been to find things to spontaneously add to the moment in a story, then find a way to link them back to whatever else is going on. If it's too strained or vague, I dump it and try again. But the more you do stuff like this, the easier it is to pull things out of a hat — any hat, any size — that work.
After all — don't we make it all up as we go along?