They estimate based on data. The biggest constraint at the moment is probably not anything that most people would think of like fuselage strain, it's the engines. The engines burn kerosene and they often run fuel rich, which leads to build up of residues (coking) in the plumbing. This is very difficult to remove and reduces the effective lifetime of the engines.

Once they get this generation of rockets settled into a pattern of high reliability reuse they'll look toward future improvements which will involve switching fuels (from kerosene to methane). LOX/Methane doesn't have the same coking problem so it could enable rockets with much longer lifetimes (a hundred or more flights per stage) and corresponding cost reductions.

Also, at that point the development cycle fundamentally changes because then you can match the airplane cycle where you build a few prototypes or test vehicles and then put them through their paces. Since you expect the vehicles to survive all the time (versus being lost routinely) you can achieve much lower development costs along with much higher reliability.