So let's say you want to check something like a new fuel nozzle.

SpaceX might design and build the nozzle, then put it in the rocket and launch it. It might work how they intended, or it might not, but they'll find out immediately. They'll make changes, build a new nozzle, launch another rocket, and continue until it works like they want.

NASA will do a lot more testing, simulation, redesigning, etc. until they KNOW that the nozzle will perform perfectly on the first try.

On the surface, NASA's approach sounds cheaper because you aren't wasting rockets. In reality it looks like SpaceX's approach might be better.

SpaceX is willing to blow up a rocket, even if it exploding is fully planned and expected. That's it, really not hard to comprehend.