Methodology is bullshit: principles for product velocity

It would be best to skip all the fluff and build exactly what customers would find valuable. IF you knew what that was. So how do you figure that out with confidence before building something?