Amazon is the one performing the experiment; if the sample size isn't big enough to be useful, that's a mark against their experimental method, not the feedback

Amazon is the one performing the experiment; if the sample size isn't big enough to be useful, that's a mark against their experimental method, not the feedback of the people on this forum. The claim that this is actually valuable in some way should be the one with the burden of proof, and it's unreasonable to argue that flaws in an experiment showing one result somehow imply that the opposite is more likely because of it.