Well, the fact is we’re looking at using toolbar data, and that’s using toolbar data only from people who have opted in. But that’s looking at real world load times from people. For example, if you’re in the United States, we might say, how long does it take to load this particular page? And so if we’re looking at that, and it takes a long time, sometimes it’s not necessarily your site. It could be the network connectivity. But it’s a good thing to bear in mind. It’s coming from all these different users, who can have dial-up lines. They can have slow connections. And so a lot of times, people say, oh, I’m just going to throw a 500 kilobyte page out there.
So that data is based primarily on toolbar data. And we’re looking at what it looks like for real users. And so if you’ve got a lot of users who are having a slow experience, then that can affect the overall rating. One thing to bear in mind, however, is that only something like one out of 100 searches is site speed such a factor that it would actually change the rankings to a noticeable degree. So that’s something on the order of one in 1,000 sites have truly site speed as a really big issue for them.
It’s always good to see if you can move a little bit faster and try to return results to users a little bit faster. It makes your website experience more fluid. It makes you users happier. There are good studies that say the return on investment is definitely worth it. But at the same time, I wouldn’t stress overly about it. But I just wanted to give you a little bit of visibility about how we compute when we think a particular site is slow in Google Webmaster Tools.
by Matt Cutts - Google's Head of Search Quality Team