I don't have an XBox nor do I play console games, but the concept of collecting data on how users actually use an application is interesting, and one that I believe is currently massively under-exploited. You could measure ease of navigation by looking at how much people blunder around looking at things without doing anything there before finding what they actually want. Knowing where users spend their time helps focus on where optimisation effort should be concentrated.
Similarly, it is possible to measure how an application performs - which parts perform well, which parts are slow. The know-how to do this has existed for many years, yet it is rarely done. There is always pressure to get code out of the door, and these measurements are very much seen as expendable 'nice-to-haves'. Even applications where specific performance requirements have to be met usually don't have the means to measure that performance built-in.
I recall doing an analysis of where an application spent its time back in the 80's. The application was an automatic test system for a telephone exchange. We found that it spent most of its time in the relay driver routines, which like the rest of the application were written in Fortran 4.
We recoded the relay driver routines in assembly language. They were 10 times faster, and lifted overall performance enough for us to achieve the performance we needed.