It isn't that simple either.
As has been discussed somewhere here before there are a number of different things to measure.
I think the number being tossed about here is the number of base pair differences. However, one might also measure the number of whole genes that are not the same. Or one could take only the base pair differences in known coding sequences (leaving out the 'junk').
One could then argue that homologuous substitutions (do I have that right? ) don't matter. If the gene is different but codes for the same thing who cares. And so on and so forth.
Let's face it what the heckenhiemer does it matter if it is 94 % or 99 %? One way of looking at is we are very, very similar to chimps at the genetic level.
Then there would be another way which will arise as we know more. A 1% difference in the right place may count a LOT! For example we may find that sequences controlling brain or language are only 89% similar.
Of course, the site referenced forgot to note that even with millions of differences it becomes kind of hard to suggest that evolution can't get from a chimp like sequence (not even counting that they have been diverging some too) to our when it only takes a few changes per generation.