Friday, September 12, 2008

People love statistics, even when they don't know how to interpret them

So I was reading this article from ComputerWorld that shows statistics about Chrome users to their site and comparing to the other 3 "most popular browsers": IE, Firefox & Safari.

They go through a lot of discussion of how it varied every day, and that during the weekend it went up, then down again and then it "stabilized" at around 5% market share, a bit lower than Safari. Sounds like a meaningful analysis? Well, just look at the graph and tell me! Look at the amount of variation that happens with "old" browsers, like IE and Firefox. They seem to keep competing for the first position, even when they have already been out for many months (actually years, because I think they aggregate all versions into one metric).

My point is, if that's the variation on normal operation, why does anybody care about .5% or even 5% changes day by day and try to conclude anything about them? Yes, Chrome got a pretty big market share quickly, but that's what you expect with anything that is released by such a big and well-known company. And that's the only thing I would take from it. Leave the rest of the conclusions for when you have month-over-month data and you can actually see cannibalization.

0 comments: