Personally, I don't find ultra-specialized databases very interesting (speaking generally here, not necessarily saying that Vertica is ultra-specialized). I basically see two reasons to use something specialized: (1) replicate data into it and run queries (i.e. an accelerator); and (2) put lots of low value data in it, that you'd otherwise throw away, and try to extract some signal from it. (There are a couple other cases, like real-time systems, but that is outside of what I ordinarily deal with.)

Pretty much. We had a specific business requirement of near real-time statistics and historical statistics available at an entity / day level of granularity for advertisers and publishers using our platform.

Vertica occupies a very sweet spot for us in that regard, but it requires a lot of specialised approaches. I've managed to break its tuple mover a couple of times when loading historical raw data in large quantities.