I then look at selling an oppositional side (in this case, the short put "wing") (1) for at least .10 more in credit than it cost me to roll the tested side; (2) with the highest probability of profit I can do that for; and (3) that does not result in an "inverted" iron condor (trust me, you do not want to try to work an inverted iron condor; an inverted short strangle is a different matter; I'll discuss that in some other post if I'm ever presented with that situation with a live trade I've posted here).
In the event that you cannot sell an oppositional side against for at least .10 ($10) more in credit than you received for rolling the tested side, consider (1) not attempting to improve the tested side at that point in time (i.e., rolling it for duration "as is"); (2) other expiries that are farther out in time; or (3) both.
Lastly, be mechanical with the way in which you "take off" rolled out setups. You can do one of two things: (a) look to take off the sides (rolled or otherwise) for near worthless as you would with any original setup (doing things this way relieves you of the burden of calculating all the credits collected and debits for a particular setup, rolls, and the sale or oppositional sides against); (b) examine your "options trade chain", total the credits and debits collected, subtract the total debits paid from the credits collected, and determine what your "scratch point" is for the original setup, rolls, etc. (i.e., where debits paid = credits received) (my options platform does this neatly for me, so I can generally determine this with a glance, thank goodness).
Naturally, there are other things you can do to "assist" your tested sides, but they usually entail additional risk (e.g., laddering out spreads in time, selling additional spreads in the same expiry, widening your spreads, etc.). I confess that I do some of those things, but I usually confine them to core position trades, such as in the index ETF's where I'm probably going to have positions on anyways, so it's not as though I'm putting on additional risk solely for the sake of working my way out of a tested side (although it can also have that welcome side effect).