The changes described in the article look good. However, were they A/B tested? While Amazon.com may look cluttered, they used A/B testing to evolve the site to what it is now.
Providing the user with a few options can be nice, but it doesn't work for all sites.
One of my companies sells gun supplies and accessories. There are tons of manufacturers and parts. Thus, for the redesign of our site, we went with a mega menu structure to help show all of the different departments and manufacturers.
People have responded very well to it.
The web site in the article is absolutely gorgeous though. I really like what they've done with the shopping cart.
I don't think this was addressed, but if so please correct me: what was the relative change per visitor?
In other words, adjusting for the increased traffic that you would expect over 5 years, do we see additional revenue growth? Or have they simply gotten more traffic?