I’m only about 11k words into my first Scrivener project, a novel. I know that certain words, phrases, sentences, ideas, events, things, skills, resources, etc. are setting the stage or laying the groundwork for important developments later in the book. I want to “track” and “come back to” this early stuff, but I’m unclear on which advanced software features are best suited to do this. I’ll give you some examples of what I mean. This is first draft stuff and may all go, but I need to know what and where it is.
A long time before the main action, a character murdered someone by stabbing him to death. I need a good way of tracking my use of words like “cut” and synonyms and reminding myself which words I have used in this way.
Some early events or perceptions will serve as early warnings, foreboding, or premonitions of later danger or threats. I will need to go back and tweak the early stuff to make it fit the later stuff better, if only for consistency or continuity, but perhaps for other reasons. I need an easy way to go back and find all the early pointers.
Some objects, knowledge, skills, and resources are introduced early so that they will not look like a deus ex machina when used or exploited later. It would be helpful to know exactly where and what the early stuff was, to ensure that I covered all the bases, and perhaps to find superfluous stuff for deletion.
I have not actually used metadata or keywords yet, but they appear to be the two most relevant features. Any opinions on which would be best for my purposes? Have I overlooked a different feature that might be more appropriate?
I’m not sure if I’d call them “advanced” but the regular old highlighters work good for stuff like this, especially for simple themes and tracking of that nature. You can use the Edit/Find/Find by Formatting... menu command to look up phrases by highlight colour, later on. It’s good for stuff like that where explanation isn’t really required. You might know a sentence is sloppy or a little weak, but don’t want to fester over it while writing, so mark it for later—that sort of thing.
For markings that require a little explanation as to why it was marked, I like inline annotations for that. I like to come up with keywords for significant things so that I can search for the keyword later on. You can put Scrivener links in them as well, so if scene a refers to scene b you can put a link there to make it easy to jump to in the future. For longer notes and other types of things, I like to use inspector comments. It depends on whether or not the marking will be useful while reading the text, really. For non-fiction I like to use comments to mark the location of figures and tables, since they can all be listed in the sidebar and clicking on one jumps to it. This strikes me as the best tool for #2, which is still at the textual level, but slightly more complex than a simple word or phrase you are tracking.
For broader things, potentially in the realm of #3, Keywords are good for that, you’re on the right track there. The nice thing about keywords over text annotation and highlighting is that you can easily gather all of the pieces of the book together that share the keyword. They can also be useful for spotting stuff that went nowhere. Maybe you write a scene and slap a new keyword on it—then kind of forget about that whole bit of plot. Later on the fact that nothing else has that keyword could be a clue that it’s worth snipping out. You can see how keywords are being exploited by turning on keyword colours on the corkboard or showing them as an outliner column.
Keywords are good when something might have several applications. If you find yourself with just a small list of keywords, say one for each plot line, and never add more than one to a particular scene—labels might be a better approach for that. But I think from your description of #3, keywords will work better for you.
Anyway, that’s all just a few basic ideas. If you want to look at what is available with a bit more depth, Chapter 10 (pg. 90) & and Ch. 17 (pg. 202) in the user manual PDF are the two places that discuss not only what can be done, but explore some usage scenarios as well.
As a brief addition to Ioa’s excellent round-up, there is also Outliner/Inspector custom meta-data.
All in all, my advice would be to try all of the suggestions; Scrivener is almost over-supplied with “meta-text layers”, and there really are no rules.