Unless they can statistically prove over a larger population that it sounds better, it don't mean jack. The ultimate approach in audio so far has been Floyd and Olive's at Harman, where they can predict with 0.86 correlation coefficient which speaker listeners will prefer in blind listening tests. And when bass extension is similar, that number goes up to 0.996. Believing anything someone or some company subjectively says without going through this kind of statistical scrutiny is akin to believing in get-rich-quick schemes. Unfortunately, the human psyche is built to want to believe in these kinds of schemes.
The exploitation of human frailty is often the marketers goal.
On the other hand folks like Floyd Toole, Sean Olive and more recently Amir Majidimehr at ASR have done some good work in ushering in a new era of measurement based "statistical scrutiny" for which an increasing number of peoplebseem to be grateful.