There was story floating around here not too long ago that had IE 6 at around 4%. Incidentally, Safari was around 4%. I know a lot of devs that want to make sure things look great in Safari but don't care about IE 6 at all. Granted the obvious argument is that 4% of Safari users is a large percentage of Mac users, but it just goes to show how deceiving browser share numbers can be.
I'd also posit that while 2% may seem like nothing for a site with a small flow of traffic, it equates to a lot of money for a high volume consumer-oriented site. I run into a lot of this through my affiliation with Mogotest. During our customer development process I was amazed to find that several of the CTOs I sat down with will support any browser with greater than 0.5% share on their site. And these are guys that have definitely done their cost/benefit analysis.
Making things look great in Safari is generally as simple as making things look great in Chrome and then maybe tweaking some fonts (due to Safari's different font rendering behaviors).
Making things look great in IE6 often involves many hours and sacrifices of chickens to dark gods with names that have lost all their vowels.
I'd also posit that while 2% may seem like nothing for a site with a small flow of traffic, it equates to a lot of money for a high volume consumer-oriented site. I run into a lot of this through my affiliation with Mogotest. During our customer development process I was amazed to find that several of the CTOs I sat down with will support any browser with greater than 0.5% share on their site. And these are guys that have definitely done their cost/benefit analysis.