Some really good tips here, already.
I am going to be very literal, CestMoiPeridot, and say that the very first thing to do is to accurately understand whether your client is making a profit from their advertising expenditure or not.
And, if this is not known, then as rj87uk suggests, it's very important to get some sort of tracking going - from a source that you are comfortable with, whether it's Google conversion tracking, Google Analytics, or a third party.
A few thoughts, to expand:
* If you know that your client is making a profit from their advertising, then this puts you in a good position in which you can take your time and experiment with 'optimizations' such as those suggested in this thread.
* In a very real sense, if your clients are currently making more than they spend, then if they spend more they are also likely to make make more - even if you change nothing. It makes a great deal of sense to try to maximize that, of course.
* If you know that your client is not making a profit from their advertising, then things become more time-sensitive. At this point it might be best to pause campaigns which your client deems to be less important, and focus on the most important ones - then apply lessons learned to the paused campaigns later on, and re-start them.
* If you don't know whether or not your client is making a profit from their advertising, then everything you might do to improve things almost amounts to guessing - and tracking of some sort becomes a must-do.
My $0.02 ;)