Well, I think that the answer will vary by person, but my personal threshold in single-player is that I try to have at least 4ish resources overall on it. Preferably more, but if I am using higher-mark command stations, or if there is a positional advantage to the planet (if it provides buffer to some more important planet, that's something I often build up as part of my core set of planets even if the resources are not too wonderful in and of themselves). And then of course capturables would also rate consideration, but assuming no capturables on a planet, that's kind of my criteria.
Another good reason to take an otherwise-worthless planet? If it is fairly unprotected and next to an enemy planet that you want supply to (and a beachhead on). Granted, that would also cause the AI on your actual target planet to go on alert, but that can be worth it. Alternatively, it works well a couple of hops away from your target, if you just need a forward base production facility to really go after the target -- without alerting the AI -- and without having to raid over such long distances as you otherwise might. That's a tactic I use against the enemy homeworlds a lot.
So -- some tangential answers, but that's pretty much how I rate planets as far as when I decide to take them.