I sincerely hope you're not making the case that testing a hypothesis for a scientific paper is not a valid reason to perform such an experiment, but improving a website is.