Thanks, Anubhav. In our UX process, we come up with a hypothesis — like greater friction leads to more sign-ups, to give an unlikely example — and then test it through usability testing. That is, we create the UX and test it qualitatively. But you’re right: in that example, we’d only be able to validate whether the design was usable, not whether there was a quantitative effect. That’d require us to ship it and run the numbers (which we also do).