What is the end state we are going for, when a single campaign is profitable, stable, and in maintenance? My guess is that it looks something like this:
10 best pictures (photos changed over time)
1 best copy
1 best lander
1 best offer
I figure we'll find the single best of everything but will still need to keep changing the photos to mitigate burnout. Is that right?
Always be split testing.
So, likely you'll drill down to the best offer, but will constantly be running variations of the 'best' lander and ad copy.
There comes a point where you give up and let it roll I guess.
You are correct that ad images will be the aspect that is generally kept fresher with a greater spread of options to mitigate burnout.
its also good to have backup offers lined up for when the top offer goes down. that's why its crucial to test a few offers in your geo/niche.