I have two levels of backlog. The first level is my curated list of tickets that are highly worth doing in the near future and is limited in size. It's currently larger than I'd like at 30-something (for a team of a little under 10), but I'm trying to get the team to focus on it more after historically neglecting it.

The second level is literally just everything else. Hundreds upon hundreds of tickets, ranging from restructuring unit tests (which will frankly never happen unless the structure of the tests somehow became a major barrier) to cool features that just aren't important enough yet (or would take too long). Plus all the super low risk bugs, often in edge cases that nobody really cares about yet or aren't worth the time to fix yet. And then there's all the automation style tickets about improving the handling of something (commonly edge cases of things already automated for the happy path), but often that something just isn't common enough to be worth it.

Tickets in the second level sometimes do get done. Usually because some issue becomes more common, enough people ask for it, or we simply finally have time for a new feature (can only do so many of those at a time). A common theme I have is I'll encounter a problem, file a ticket, then eventually encounter the problem enough times that I go, "fuck it, I'll do it myself".