Without a link describing the methods used to garner this information and the raw data itself, these conclusions are, at best, suspect.
Their decision to create lots of "unofficial" support sites (e.g. http://getsatisfaction.com/gm - note that the "unofficial" is easy to miss) was also... controversial. Especially since companies had to pay quite a bit to take over this "support" forum.
I didn't see anything in the numbers that was a big surprise, but I don't trust the 68% leave because of the treatment they receive vs. 14% that leave due to dissatisfaction with the product or service. The numbers seem skewed toward GetSatisfaction's forte.