Even once they get the landing perfected, my concern would be the wear-and-tear of even a single launch of the rocket. It has to reduce the success probability of the next launch by some percentage.
You can sell the subsequent launches more cheaply and use them for payloads that are more expendable.
I believe the plan is that the first launch of a new first stage would be used for manned missions (once they get to that point) and then subsequent launches would be used for satellites and such. Presumably going for cheaper and cheaper satellites as it ages.
If we really are going to start launching manned missions to Mars then most of the payloads launched from earth are going to just be propellant. If you have a failure with a used rocket that's just carrying propellant as its payload, then you're pretty much out just the cost of the propellant. This is probably oversimplifying things a bit, but still.
True, but that's something you can largely evaluate on the ground by re-lighting engines in a test rig. Also, even a couple of launches with the same vehicle would dramatically reduce costs per launch.
