Algol 60 had a number of obvious deficiencies, which needed to be addressed. Both Algol 68 and CPL were designed to fix these issues, but had a level of complexity that made them impractical to implement.
Pascal, which Predates C slightly, grew out of of Wirth's much simpler (compared to Algol 68) set of changes to Algol 60.
The CPL people were having trouble with implementation, so Richards created a stripped down version (BCPL). Thompson apparently didn't have access to the full specification, but samples of it greatly informed the the design of B, which eventually begat C. Apparently the "killer-feature" of C over B at the beginning was byte-addressing, which is arguably no more or less niche than what any of these new languages claim.
Almost memory safe systems programming languages (not fully safe due to use-after-free) were already a thing since 1961.
In fact ESPOL was probably the very first systems programming language with unsafe code blocks.
That's not the category I'm talking about, though. C (and I believe Ruby) was a major shift from what came before. Those two are better compared to Clojure or Go: dramatically different value-propositions that can make it worth the time and effort to take a chance on a new, green technology. What I'm talking about is languages that are "like X but with improvements to Y and Z".