Yeah there were so many problems with 1010Data's approach and presentation that even the Register -- not known to be a great fan of Apple -- pretty much declared the difference meaningless. I figured I'd leave the Apple-side debunking to you. :-)
You know I will rip the misuse of statistical methodology, too. . . which is what they are doing here. I think that the concept that Microsoft has is flawed. Minimal battery and minimal graphics capability in the tablet section is not a good idea, and for maximum functionality, you have to have a surface to rest the Surface on and make it into a laptop at minimum.
This is one of the statistical methodologies I have problem with . . . Trends should not be shown by discrete points, but be lines over time. This is a set of points connected by meaningless, arbitrary lines intended to obfuscate the information by implying the trend is NOT discrete points. They've also represented the data as Year To Date, safely excluding last year's holiday season. Why? Because there would have been a HUGE spike for both Apple and Amazon swamping Microsoft. This is a classic move to make one set of data look better (usually the one paying for the study) look better.