"Related" is a often a non-descript "hold all" for a range of different types of relationships - the key works that a paper cites (added as you read the paper), and other key works that cites it, maybe other important work by the same authors, and any other relevant related work. I wish I could use it more specifically, but that wouldn't be to connect items thematically (at least not in its current form). And the Reference plugin actually gives me all the cited papers, as well as all the papers that have cited the paper, in the right pane. So I would prefer not to use Related for those (although the plugin does have an option to add individual extracted cited papers to Related). But that plugin is still being refined; it doesn't always work as well as it could. Also, I think Zotero needs its *own* two specific right pane sections for the key Cited and Cited-By paper categories (that metadata is now readily retrievable now from online repositories). Which would then leave Related just for related works that don't fall into those two key categories. Until then, Related remains a bit of a dumpster ... and as noted above, it has no way for the user to indicate *why* an item has been placed under that heading (a key deficiency). It also only shows titles. So if you are re-reading a paper that cites by author, it is often difficult to tell if you already have added a particular citation under Related (Related could perhaps be made to show the full author-title-date on hovering over items in its list).

Each of the options mentioned - tags, collections, Related (also add saved searches) - best suits a particular way of thinking about and retrieving grouped items. How you think about your knowledge domain is probably unique to you.

I try to apply all the appropriate subject tags for each item as they are added to my library, or soon after (I devised my complete tag list very early on; adding new tags to that list later is harder ... to go back and apply them to existing items). My tags vary in breadth - from the board topic to pretty specific technical aspects. All the items that end up with a particular subject tag can include a lot of items that might not be the best work (on most of my key topics of interest I try to add everything ever published on the subject to my library). But tags do often allow me to pull items from my library that I have forgotten about. I have been heavily using tags since the first reference managers decades ago. A good personal tag system is very powerful.

I create "topic" subcollections as I need them, every time I need a collection of the best work on a particular subject ... to answer a question in my mind, or for something I am writing. I put items in that subcollection based firstly on my memory of which are the best works, and maybe other good work from what the relevant tag(s) search throws up. And then as I read new work as it comes out, I try to remember to add the best new works to any of the existing relevant topic subcollections. But I am wary of collection "bloat" - I don't create a new subcollection until I really need it. I haven't got to the point of archiving collections that have passed their usefulness, but I may do so at some point.

I have dozens of saved searches - for key authors or labs, and a few key topic tags (or set of tags) that I find myself needing often. I don't use quick searches very often.