In the press and on the bulletin boards, the graphic designers tend to take a beating. And not surprisingly. They are the inarticulate ones, remember? Thus, you get articles that malign innovative designers without rightly discerning the purpose of their sitesor, worse, you get outright dreck written by blind guides who wouldn't know aesthetic appeal if it stripped bare and gave them a table dance.
But, waitthe usability experts have their statistics! It has been documented! The users are on their side! But what questions are these usability studies asking? "Could you find it?" "Were you able to accomplish it?" "Articulate to us in words what you were able to do." Such questions presuppose a "Martian" criterion for user experience. If the site is a "Martian" site (logical, rational, left-brained), then it will score well on Neilsen's usability test. If the site is a "Venusian" site (intuitive, emotional, right-brained), then it will score poorly on Neilsen's usability test. Were Neilsen to ask questions like, "What adjective best expresses how this site made you feel?" he would get dramatically different statistical results. But, of course, questions like that would be touchy-feely and unscientific. To use his own jargon, then, Neilsen-esque usability testing is, in the overall scheme of things, yet another form of "voodoo" usability. It finds what it is looking for and ignores what it is unable to measure.
But what if Levis doesn't have a huge database of products that it's selling online? What if Levis just wants people to feel that its vintage clothing is rugged and somewhat extreme? What if a successful site to Levis is a site that conveys an emotion, an attitude, a world view? In other words, what if Levis is branding? Most usability tests are impotent to evaluate the success of a site in terms of conveyed emotion because emotion is something that most users (and most humans) have difficulty articulating, particularly in response to multiple-choice questions. But just because a positive interactive experience can't be charted doesn't mean that it hasn't occurred.