There’s already a lot of great advice here (Merlin’s guide, focus on Flex machines, try some simple scene changes etc.) but one overall thing I’d suggest is to just go wild with experimentation.
When I’m trying to figure out something new on the Octa, it sometimes helps me to just make a complete mess. I’ll create a blank project, start building a Part and trying combinations of machines/effects/etc., and then I might reach a point where I don’t really like where it’s going. So I’ll just jump to a new Bank and start fresh again, incorporating what I learned on the first go and trying something a little different.
Eventually, after multiple experiments, I will have learned a ton and honed my approach. Even if I end up going back to the original idea, the time I spent experimenting is not wasted because I picked up a lot of information along the way. Some of the subtle yet important stuff in the Octa is just easiest to learn by putting your hands on it and giving it a shot.
This might not seem like the most efficient or organized way to build a set or grow your knowledge, but personally I just like working quickly and not worrying too much about where I’m headed while I’m in the middle of figuring things out. I prefer to go nuts and learn by doing, and the Octa is an awesome environment for that.